Chain-of-Thought (CoT) reasoning is a critical capability for large language models (LLMs), enabling them to tackle complex multi-step tasks. While base LLMs, pre-trained on general text corpora, often struggle with reasoning due to a lack of specialized training, recent studies reveal their latent reasoning potential tied to hidden states. However, existing hidden state manipulation methods, such as linear activation steering, suffer from limitations due to their rigid and unconstrained nature, often leading to distribution shifts and degraded text quality. In this work, we propose a novel approach for eliciting CoT reasoning from base LLMs through hidden state manipulation grounded in probabilistic conditional generation. By reformulating the challenge as an optimization problem with a balanced likelihood and prior regularization framework, our method guides hidden states toward reasoning-oriented trajectories while preserving linguistic coherence. Extensive evaluations across mathematical, commonsense, and logical reasoning benchmarks demonstrate that our approach consistently outperforms existing steering methods, offering a theoretically principled and effective solution for enhancing reasoning capabilities in base LLMs.

With the growing potential of large language models (LLMs) in the legal domain, domain-specific finetuning and retrieval-augmented generation (RAG) methods have received widespread attention. However, current methods still suffer from hallucination risk and failing to resolve semantic drift and adapt to varying citation numbers. To address this, we propose Authoritative and Accurate Lawyer (AALawyer), a complementary dual-retriever framework based on the Legal Syllogism and the nature of different legal data. First, we introduce Symbolic Constrained Retrieval (SCR) for closed-set article retrieval, by constraining retrieval to the generative prediction. Second, we build Analogical Precedent Retrieval (APR) to retrieve open-set judicial precedents for reasoning with a newly collected large criminal dataset.Extensive experiments, including LawBench, our Hallucination Risk-Benchmark, and comprehensive ablation studies, demonstrate the effectiveness of AALawyer, which mitigates hallucinations while improving the explainability of legal reasoning.

The spectacular success of training large models on extensive datasets highlights the potential of scaling up for exceptional performance. To deploy these models on edge devices, knowledge distillation (KD) is commonly used to create a compact model from a larger, pretrained teacher model. However, as models and datasets rapidly scale up in practical applications, it is crucial to consider the applicability of existing KD approaches originally designed for limited-capacity architectures and small-scale datasets. In this paper, we revisit current KD methods and identify the presence of a small-data pitfall, where most modifications to vanilla KD prove ineffective on large-scale datasets. To guide the design of consistently effective KD methods across different data scales, we conduct a meticulous evaluation of the knowledge transfer process. Our findings reveal that incorporating more useful information is crucial for achieving consistently effective KD methods, while modifications in loss functions show relatively less significance. In light of this, we present a paradigmatic example that combines vanilla KD with deep supervision, incorporating additional information into the student during distillation. This approach surpasses almost all recent KD methods. We believe our study will offer valuable insights to guide the community in navigating beyond the small-data pitfall and toward consistently effective KD.

Marine Saliency Segmentation (MSS) plays a pivotal role in a wide range of vision-based marine exploration tasks. However, existing techniques often face the dilemma of imprecise boundaries due to the interference-rich nature of underwater environments, where suspended particles, low contrast, and color distortion hinder accurate segmentation. Although diffusion models have shown impressive performance in visual tasks, their potential to incorporate contextual semantics for enhancing feature learning of region-level salient objects remains underexplored, thereby hindering segmentation outcomes. Building on this insight, we propose DiffMSS, a novel marine saliency segmenter based on the diffusion model, which utilizes semantic knowledge distillation to guide the detection of marine salient objects. Specifically, we design the Word-level Semantic Saliency Extraction module that identifies salient terms at the word level from the captions by computing region-word similarity. These high-level semantic features are distilled into the Conditional Feature Learning Network to generate accurate and semantically informed diffusion conditions. The Object-Focused Conditional Diffusion module then leverages these conditions to iteratively generate fine-grained segmentation masks of marine instances, while a Consensus Deterministic Sampling scheme is further employed to suppress overconfident mis-segmentations and enhance structural fidelity. Extensive experiments demonstrate the superior performance of DiffMSS over state-of-the-art methods in both quantitative and qualitative evaluations. Our code and pre-trained models will be released on GitHub.

Generating motion sequences conforming to a target style while adhering to the given content prompts requires accommodating both the content and style. In existing methods, the information usually only flows from style to content, which may cause conflict between the style and content, harming the integration. Differently, in this work we build a bidirectional control flow between the style and the content, also adjusting the style towards the content, in which case the style-content collision is alleviated and the dynamics of the style is better preserved in the integration. Moreover, we extend the stylized motion generation from one modality, i.e. the style motion, to multiple modalities including texts and images through contrastive learning, leading to flexible style control on the motion generation. To further boost the performance, we advance the motion diffusion to motion-aligned temporal latent diffusion by developing a novel motion VAE. Extensive experiments demonstrate that our method significantly outperforms previous methods across different datasets, while also enabling multimodal signals control. The code of our method will be made publicly available.

Underwater imaging is often plagued by significant degradation in visual quality, primarily due to the effects of light absorption and scattering in water. Although recent underwater image enhancement (UIE) methods rely on the current advances in deep neural network architecture designs, there is still considerable room for improvement in cross-scene robustness and computational efficiency. Diffusion models have shown great success in image generation, prompting us to explore their application to UIE tasks. However, directly applying them to UIE tasks will pose two challenges,i.e., high computational budget and color unbalanced perturbations. To tackle these issues, we propose DiffColor, a distribution-aware diffusion and cross-spectral refinement model for efficient UIE. Unlike single-noise image restoration tasks, underwater imaging exhibits unbalanced channel distributions due to the selective absorption of light by water. To address this, we design the Global Color Correction to balance the diverse color shifts, thereby avoiding potential global degradation disturbances during the denoising process. Instead of diffusing in the raw pixel space, we transform the image into the wavelet domain to obtain such low-frequency and high-frequency spectra. For the sacrificed image details caused by underwater scattering, we further present the Cross-Spectral Detail Refinement to enhance the high-frequency details, which are then integrated with the low-frequency signal as a dual-condition for guiding the diffusion. This strategy ensures the high-fidelity of sampled content and compensates for the sacrificed details. Extensive experiments demonstrate the superior performance of DiffColor over state-of-the-art methods in both quantitative and qualitative evaluations.

Underwater salient object detection (USOD) plays a pivotal role in various vision-based marine exploration tasks. However, existing USOD techniques face the dilemma of object mislocalization and imprecise boundaries due to the complex underwater environment. The quality degradation of raw underwater images (caused by selective absorption and medium scattering) makes it challenging to perform instance detection directly. One conceivable approach involves initially removing visual disturbances through underwater image enhancement (UIE), followed by saliency detection. However, this two-stage approach neglects the potential positive impact of the restoration procedure on saliency detection due to it executes in a cascade. Based on this insight, we propose a generalized prior-involved diffusion model, called WaterDiffusion for collaborative underwater saliency detection and visual restoration. Specifically, we first propose a revised self-attention joint diffusion, which embeds dynamic saliency masks into the diffusive network as latent features. By extending the underwater degradation prior into the multi-scale decoder, we innovatively exploit optical transmission maps to aid in localizing underwater salient objects. Then, we further design a gate-guided binary indicator to select either normalized or raw channels for improving feature generalization. Finally, the Half-quadratic Splitting is introduced into the unfolding sampling to refine saliency masks iteratively. Comprehensive experiments demonstrate the superior performance of WaterDiffusion over state-of-the-art methods in both quantitative and qualitative evaluations.

Diffusion models (DMs) have demonstrated great potential in the field of adversarial robustness, where DM-based defense methods can achieve superior defense capability without adversarial training. However, they all require huge computational costs due to the usage of large-scale pre-trained DMs, making it difficult to conduct full evaluation under strong attacks and compare with traditional CNN-based methods. Simply reducing the network size and timesteps in DMs could significantly harm the image generation quality, which invalidates previous frameworks. To alleviate this issue, we redesign the diffusion framework from generating high-quality images to predicting distinguishable image labels. Specifically, we employ an image translation framework to learn many-to-one mapping from input samples to designed orthogonal image labels. Based on this framework, we introduce an efficient Image-to-Image diffusion classifier with a pruned U-Net structure and reduced diffusion timesteps. Besides the framework, we redesign the optimization objective of DMs to fit the target of image classification, where a new classification loss is incorporated in the DM-based image translation framework to distinguish the generated label from those of other classes. We conduct sufficient evaluations of the proposed classifier under various attacks on popular benchmarks. Extensive experiments show that our method achieves better adversarial robustness with fewer computational costs than DM-based and CNN-based methods.

Prior efforts in light-weight model development mainly centered on CNN and Transformer-based designs yet faced persistent challenges. CNNs adept at local feature extraction compromise resolution while Transformers offer global reach but escalate computational demands O(N^2). This ongoing trade-off between accuracy and efficiency remains a significant hurdle. Recently, state space models (SSMs), such as Mamba, have shown outstanding performance and competitiveness in various tasks such as language modeling and computer vision, while reducing the time complexity of global information extraction to O(N). Inspired by this, this work proposes to explore the potential of visual state space models in light-weight model design and introduce a novel efficient model variant dubbed EfficientVMamba. Concretely, our EfficientVMamba integrates a atrous-based selective scan approach by efficient skip sampling, constituting building blocks designed to harness both global and local representational features. Additionally, we investigate the integration between SSM blocks and convolutions, and introduce an efficient visual state space block combined with an additional convolution branch, which further elevate the model performance. Experimental results show that, EfficientVMamba scales down the computational complexity while yields competitive results across a variety of vision tasks. For example, our EfficientVMamba-S with 1.3G FLOPs improves Vim-Ti with 1.5G FLOPs by a large margin of 5.6% accuracy on ImageNet.

Deep neural networks (DNNs) have achieved significant success across various tasks, but ensuring reliable uncertainty estimates, known as model calibration, is crucial for their safe and effective deployment. Modern DNNs often suffer from overconfidence, leading to miscalibration. We propose a novel post-hoc calibration method called feature clipping (FC) to address this issue. FC involves clipping feature values to a specified threshold, effectively increasing entropy in high calibration error samples while maintaining the information in low calibration error samples. This process reduces the overconfidence in predictions, improving the overall calibration of the model. Our extensive experiments on datasets such as CIFAR-10, CIFAR-100, and ImageNet, and models including CNNs and transformers, demonstrate that FC consistently enhances calibration performance. Additionally, we provide a theoretical analysis that validates the effectiveness of our method. As the first calibration technique based on feature modification, feature clipping offers a novel approach to improving model calibration, showing significant improvements over both post-hoc and train-time calibration methods and pioneering a new avenue for feature-based model calibration.

Text-to-image diffusion models have demonstrated remarkable capabilities in creating images highly aligned with user prompts, yet their proclivity for memorizing training set images has sparked concerns about the originality of the generated images and privacy issues, potentially leading to legal complications for both model owners and users, particularly when the memorized images contain proprietary content. Although methods to mitigate these issues have been suggested, enhancing privacy often results in a significant decrease in the utility of the outputs, as indicated by text-alignment scores. To bridge the research gap, we introduce a novel method, PRSS, which refines the classifier-free guidance approach in diffusion models by integrating prompt re-anchoring (PR) to improve privacy and incorporating semantic prompt search (SS) to enhance utility. Extensive experiments across various privacy levels demonstrate that our approach consistently improves the privacyutility trade-off, establishing a new state-of-the-art.

Model calibration is essential for ensuring that the predictions of deep neural networks accurately reflect true probabilities in real-world classification tasks. However, deep networks often produce over-confident or under-confident predictions, leading to miscalibration. Various methods have been proposed to address this issue by designing effective loss functions for calibration, such as focal loss. In this paper, we analyze its effectiveness and provide a unified loss framework of focal loss and its variants, where we mainly attribute their superiority in model calibration to the loss weighting factor that estimates sample-wise uncertainty. Based on our analysis, existing loss functions fail to achieve optimal calibration performance due to two main issues: including misalignment during optimization and insufficient precision in uncertainty estimation. Specifically, focal loss cannot align sample uncertainty with gradient scaling and the single logit cannot indicate the uncertainty. To address these issues, we reformulate the optimization from the perspective of gradients, which focuses on uncertain samples. Meanwhile, we propose using the Brier Score as the loss weight factor, which provides a more accurate uncertainty estimation via all the logits. Extensive experiments on various models and datasets demonstrate that our method achieves state-of-the-art (SOTA) performance.1

We introduce UniGraspTransformer, a universal Transformer-based network for dexterous robotic grasping that simplifies training while enhancing scalability and performance. Unlike prior methods such as UniDexGrasp++, which require complex, multi-step training pipelines, UniGraspTransformer follows a streamlined process: first, dedicated policy networks are trained for individual objects using reinforcement learning to generate successful grasp trajectories; then, these trajectories are distilled into a single, universal network. Our approach enables UniGraspTransformer to scale effectively, incorporating up to 12 self-attention blocks for handling thousands of objects with diverse poses. Additionally, it generalizes well to both idealized and real-world inputs, evaluated in state-based and vision-based settings. Notably, UniGraspTransformer generates a broader range of grasping poses for objects in various shapes and orientations, resulting in more diverse grasp strategies. Experimental results demonstrate significant improvements over state-of-the-art, UniDexGrasp++, across various object categories, achieving success rate gains of 3.5%, 7.7%, and 10.1% on seen objects, unseen objects within seen categories, and completely unseen objects, respectively, in the vision-based setting. Project page: https://dexhand.github.io/UniGraspTransformer/.

Traditional video instance segmentation (VIS) models rely on extensive per-frame video annotations, which are both time-consuming and costly. In this paper, we present MinMaxVIS, a novel VIS framework that reduces the dependency on fully labeled video datasets by utilizing a small set of labeled images from the target domain along with a large volume of general-domain, unlabeled images. MinMaxVIS operates in three stages: first, a preliminary segmentation model is trained on the small labeled set from the target domain; this model then retrieves relevant instances from the unlabeled dataset to build a high-quality pseudo-labeled set, ensuring a rich content alignment with the target domain while avoiding the inefficiencies of large-scale semi-supervised learning across the entire unlabeled set. Finally, we train MinMaxVIS on a combination of labeled and pseudo-labeled data, addressing challenges such as noise in pseudo-labels and instance association across frames. To simulate object continuity, we augment static images to create paired frames, allowing MinMaxVIS to capture instance associations effectively. MinMaxVIS outperforms the prior image-driven approach, MinVIS, achieving superior mAP scores with significantly reduced labeled data. For instance, MinMaxVIS with a Swin-L backbone attains 62.2 mAP on YouTube-VIS 2019 using only 2% labeled data and additional unlabeled images from SA-1B. This surpasses MinVIS, which uses the same backbone trained on fully labeled YouTube-VIS 2019, by 0.6 mAP.

This paper introduces ThoughtProbe, a novel inference-time framework that leverages the hidden reasoning features of Large Language Models (LLMs) to improve their reasoning performance.Unlike previous works that manipulate the hidden representations to steer LLM generation, we harness them as discriminative signals to guide the tree-structured response space exploration.In each node expansion, a classifier serves as a scoring and ranking mechanism that efficiently allocates computational resources by prioritizing higher score candidates for continuation.After completing the tree expansion, we collect answers from all branches to form a candidate answer pool.We then propose a branch-aggregation method that marginalizes over all supporting branches by aggregating their CoT scores, thereby identifying the optimal answer from the pool.Experimental results show that our framework's comprehensive exploration not only covers valid reasoning chains but also effectively identifies them, achieving significant improvements across multiple arithmetic reasoning benchmarks.The code is available at https:// github.com/Zijian007/Thoughtprobe.

We introduce Lumina-Image 2.0, an advanced text-to-image generation framework that achieves significant progress compared to previous work, Lumina-Next. Lumina-Image 2.0 is built upon two key principles: (1) Unification - it adopts a unified architecture (Unified Next-DiT) that treats text and image tokens as a joint sequence, enabling natural cross-modal interactions and allowing seamless task expansion. Besides, since high-quality captioners can provide semantically well-aligned text-image training pairs, we introduce a unified captioning system, Unified Captioner (UniCap), specifically designed for T2I generation tasks. UniCap excels at generating comprehensive and accurate captions, accelerating convergence and enhancing prompt adherence. (2) Efficiency - to improve the efficiency of our proposed model, we develop multi-stage progressive training strategies and introduce inference acceleration techniques without compromising image quality. Extensive evaluations on academic benchmarks and public text-to-image arenas show that Lumina-Image 2.0 delivers strong performances even with only 2.6B parameters, highlighting its scalability and design efficiency. We have released our training details, code, and models at https://github.com/Alpha-VLLM/Lumina-Image-2.0.

Masked autoencoders (MAEs) represent a prominent self-supervised learning paradigm in computer vision. Despite their empirical success, the underlying mechanisms of MAEs remain insufficiently understood. Recent studies have attempted to elucidate the functioning of MAEs through contrastive learning and feature representation analysis, yet these approaches often provide only implicit insights. In this paper, we propose a new perspective for understanding MAEs by leveraging the information bottleneck principle in information theory. Our theoretical analyses reveal that optimizing the latent features to balance relevant and irrelevant information is key to improving MAE performance. Building upon our proofs, we introduce MI-MAE, a novel method that optimizes MAEs through mutual information maximization and minimization. By enhancing latent features to retain maximal relevant information between them and the output, and minimizing irrelevant information between them and the input, our approach achieves better performance. Extensive experiments on standard benchmarks show that MI-MAE significantly outperforms MAE models in tasks such as image classification, object detection, and semantic segmentation. Our findings validate the theoretical framework and highlight the practical advantages of applying the information bottleneck principle to MAEs, offering deeper insights for developing more powerful self-supervised learning models.

Text-to-image diffusion models have achieved unprecedented proficiency in generating realistic images. However, their inherent tendency to memorize and replicate training data during inference raises significant concerns, including potential copyright infringement. In response, various methods have been proposed to evaluate, detect, and mitigate memorization. Our analysis reveals that existing approaches significantly underperform in handling local memorization, where only specific image regions are memorized, compared to global memorization, where the entire image is replicated. Also, they cannot locate the local memorization regions, making it hard to investigate locally. To address these, we identify a novel "bright ending" (BE) anomaly in diffusion models prone to memorizing training images. BE refers to a distinct cross-attention pattern observed in text-to-image diffusion models, where memorized image patches exhibit significantly greater attention to the final text token during the last inference step than non-memorized patches. This pattern highlights regions where the generated image replicates training data and enables efficient localization of memorized regions. Equipped with this, we propose a simple yet effective method to integrate BE into existing frameworks, significantly improving their performance by narrowing the performance gap caused by local memorization. Our results not only validate the successful execution of the new localization task but also establish new state-of-the-art performance across all existing tasks, underscoring the significance of the BE phenomenon.

The diffusion sampling process faces a persistent challenge stemming from its incoherence, attributable to varying noise directions across different timesteps. Our Representative Guidance (RepG) offers a new perspective to address this issue by reformulating the sampling process with a coherent direction toward a representative target. From this perspective, classic classifier guidance reveals its drawback in lacking meaningful representative information, as the features it relies on are optimized for discrimination and tend to highlight only a narrow set of class-specific cues. This focus often sacrifices diversity and increases the risk of adversarial generation. In contrast, we leverage self-supervised representations as the coherent target and treat sampling as a downstream task—one that focuses on refining image details and correcting generation errors, rather than settling for oversimplified outputs. Our Representative Guidance achieves superior performance and demonstrates the potential of pre-trained self-supervised models in guiding diffusion sampling. Our findings show that RepG not only significantly improves vanilla diffusion sampling, but also surpasses state-of-the-art benchmarks when combined with classifier-free guidance.

As diffusion models become increasingly popular, the misuse of copyrighted and private images has emerged as a major concern. One promising solution to mitigate this issue is identifying the contribution of specific training samples in generative models, a process known as data attribution. Existing data attribution methods for diffusion models typically quantify the contribution of a training sample by evaluating the change in diffusion loss when the sample is included or excluded from the training process. However, we argue that the direct usage of diffusion loss cannot represent such a contribution accurately due to the calculation of diffusion loss. Specifically, these approaches measure the divergence between predicted and ground truth distributions, which leads to an indirect comparison between the predicted distributions and cannot represent the variances between model behaviors. To address these issues, we aim to measure the direct comparison between predicted distributions with an attribution score to analyse the training sample importance, which is achieved by Diffusion Attribution Score (\textit{DAS}). Underpinned by rigorous theoretical analysis, we elucidate the effectiveness of DAS. Additionally, we explore strategies to accelerate DAS calculations, facilitating its application to large-scale diffusion models. Our extensive experiments across various datasets and diffusion models demonstrate that DAS significantly surpasses previous benchmarks in terms of the linear data-modelling score, establishing new state-of-the-art performance.

Large Vision-Language Models (LVLMs) exhibit impressive capabilities in complex visual tasks but are prone to hallucination, especially in open-ended generation tasks. This paper explores why LVLMs tend to hallucinate and how to mitigate it. First, we conduct causal mediation analysis through counterfactual edits on specific modules in LVLMs. Our results disclose that Multi-Head Attention (MHA) modules contribute more to the probability of generating hallucination words than multi-layer perceptron modules. We then identify specific heads that are responsible for hallucination, referred to as hallucination heads. Second, we examine the behavior of hallucination heads. We find that they are concentrated in the middle and deeper layers, displaying a strong attention bias toward text tokens. Further, we show that the attention patterns of certain hallucination heads exhibit greater similarity to the base language model and change slowly during the instruction tuning process. Finally, we propose two simple yet effective methods to mitigate hallucination: one is training-free and can be applied directly during decoding, while the other involves fine-tuning. Both methods are targeted for hallucination heads to reduce their reliance on text tokens. Notably, our methods achieve up to 1.7x reduction in hallucination rate for the LLaVA-v1.5-7B model in COCO captioning task, outperforming existing baselines. Overall, our findings suggest that hallucinations in LVLMs are likely to stem from certain modules, and targeted interventions can effectively mitigate these issues.

Diffusion models (DMs) have achieved record-breaking performance in image generation tasks. Nevertheless, in practice, the training-sampling discrepancy, caused by score estimation error and discretization error, limits the modeling ability of DMs, a phenomenon known as exposure bias. To alleviate such exposure bias and further improve the generative performance, we put forward a prompt learning framework built upon a lightweight prompt prediction model. Concretely, our model learns an anti-bias prompt for the generated sample at each sampling step, aiming to compensate for the exposure bias that arises. Following this design philosophy, our framework rectifies the sampling trajectory to match the training trajectory, thereby reducing the divergence between the target data distribution and the modeling distribution. To train the prompt prediction model, we simulate exposure bias by constructing training data and introduce a time-dependent weighting function for optimization. Empirical results on various DMs demonstrate the superiority of our prompt learning framework across three benchmark datasets. Importantly, the optimized prompt prediction model effectively improves image quality with only a 5\% increase in sampling overhead, which remains negligible.

Diffusion models have demonstrated remarkable efficacy in various generative tasks with the predictive prowess of denoising model. Currently, diffusion models employ a uniform denoising model across all timesteps. However, the inherent variations in data distributions at different timesteps lead to conflicts during training, constraining the potential of diffusion models. To address this challenge, we propose a novel two-stage divide-and-conquer training strategy termed TDC Training. It groups timesteps based on task similarity and difficulty, assigning highly customized denoising models to each group, thereby enhancing the performance of diffusion models. While two-stage training avoids the need to train each model separately, the total training cost is even lower than training a single unified denoising model. Additionally, we introduce Proxy-based Pruning to further customize the denoising models. This method transforms the pruning problem of diffusion models into a multi-round decision-making problem, enabling precise pruning of diffusion models. Our experiments validate the effectiveness of TDC Training, demonstrating improvements in FID of 1.5 on ImageNet64 compared to original IDDPM, while saving about 20% of computational resources.

Group fairness based on adversarial training has gained significant attention on graph data, which was implemented by masking sensitive attributes to generate fair feature views. However, existing models suffer from training instability due to uncertainty of the generated masks and the trade-off between fairness and utility. In this work, we propose a stable fair Graph Neural Network (SFG) to maintain training stability while preserving accuracy and fairness performance. Specifically, we first theoretically derive a tight upper Lipschitz bound to control the stability of existing adversarial-based models and employ a stochastic projected subgradient algorithm to constrain the bound, which operates in a block-coordinate manner. Additionally, we construct the uncertainty set to train the model, which can prevent unstable training by dropping some overfitting nodes caused by chasing fairness. Extensive experiments conducted on three real-world datasets demonstrate that SFG is stable and outperforms other state-of-the-art adversarial-based methods in terms of both fairness and utility performance. Codes are available at https://github.com/sh-qiangchen/SFG.

Model calibration seeks to ensure that models produce confidence scores that accurately reflect the true likelihood of their predictions being correct. However, existing calibration approaches are fundamentally tied to datasets of one-hot labels implicitly assuming full certainty in all the annotations. Such datasets are effective for classification but provides insufficient knowledge of uncertainty for model calibration, necessitating the curation of datasets with numerically rich ground-truth confidence values. However, due to the scarcity of uncertain visual examples, such samples are not easily available as real datasets. In this paper, we introduce calibration-aware data augmentation to create synthetic datasets of diverse samples and their ground-truth uncertainty. Specifically, we present **Calibration-aware Semantic Mixing (CSM)**, a novel framework that generates training samples with mixed class characteristics and annotates them with distinct confidence scores via diffusion models. Based on this framework, we propose calibrated reannotation to tackle the misalignment between the annotated confidence score and the mixing ratio during the diffusion reverse process. Besides, we explore the loss functions that better fit the new data representation paradigm. Experimental results demonstrate that CSM achieves superior calibration compared to the state-of-the-art calibration approaches. Our code is [available here](https://github.com/E-Galois/CSM).

Random defense represents a promising strategy to protect neural networks from adversarial attacks. Most of these methods enhance robustness by injecting randomness into the data, increasing uncertainty for attackers. However, this randomness could reduce the generalization capacity of defense, as defense performance could be sensitive to the hyperparameters of noise added to the data, making it difficult to generalize across different datasets. Additionally, the involvement of randomness always comes with a reduction of natural accuracy, which leads to a delicate trade-off between them, which is seldom studied in random defense. In this work, we propose incorporating randomness into the network structure instead of data input by designing stochastic deformable convolution, where a random mask replaces the convolutional offset. This process promotes data independence, enhancing generalization across datasets. To study the trade-off, we conduct a theoretical analysis of both robust and clean accuracy, from a perspective of gradient cosine similarity and natural inference. Based on the analysis, we reformulate the adversarial training in our random defense framework. Extensive experiments show that our method achieves SOTA adversarial robustness and clean accuracy compared with other random defense methods.

Multimodal image fusion and semantic segmentation are critical for autonomous driving. Despite advancements, current models often struggle with segmenting densely packed elements due to a lack of comprehensive fusion features for guidance during training. While the Segment Anything Model (SAM) allows precise control during fine-tuning through its flexible prompting encoder, its potential remains largely unexplored in the context of multimodal segmentation for natural images. In this paper, we introduce SAM into multimodal image segmentation for the first time, proposing a novel framework that combines Latent Space Token Generation (LSTG) and Fusion Mask Prompting (FMP) modules. This approach transforms the training methodology for multimodal segmentation from a traditional black-box approach to a controllable, prompt-based mechanism. Specifically, we obtain latent space features for both modalities through vector quantization and embed them into a cross-attention-based inter-domain fusion module to establish long-range dependencies between modalities. We then use these comprehensive fusion features as prompts to guide precise pixel-level segmentation. Extensive experiments on multiple public datasets demonstrate that our method significantly outperforms SAM and SAM2 in multimodal autonomous driving scenarios, achieving an average improvement of 4.1% over the state-of-the-art method in segmentation mIoU, and the performance is also optimized in other multi-modal visual scenes.

Dynamic novel view synthesis (NVS) aims to render time-varying scenes from arbitrary viewpoints, balancing rendering quality and computational efficiency. While recent 4D Gaussian Splatting approaches offer promising real-time performance, they fundamentally overlook critical interdependence between Gaussians by modeling deformations independently. Our information-theoretic analysis reveals substantial mutual information across the Gaussian field, manifesting as appearance-preserving radiance coherence and motion-consistent deformation propagation. This finding establishes that rendering quality emerges from coordinated transformation rather than independent processing. We propose Correlation-aware Dynamic Gaussian Splatting (CaDGS) with our novel Gaussian Correlation Tensor Projection (GCTP) method, which efficiently transforms the complex O(n3) mutual information tensor into a dual-channel O(n2) spatial matrix, preserving the critical topological structure of Gaussian interactions. Combined with our Spatio-Temporal Deformation Consistency (STDC) learning, which enforces volumetric coherence through tensor-guided regularization across multiple scales, CaDGS prevents geometric distortions and texture inconsistencies common in previous approaches. Experimental results demonstrate state-of-the-art performance, achieving 32.4 PSNR on the Neu3D dataset with fewer Gaussians while maintaining rendering speeds of 323 FPS at 1353 × 1014 resolution.

Recent advancements in graph unlearning models have enhanced model utility by preserving the node representation essentially invariant, while using gradient ascent on the forget set to achieve unlearning. However, this approach causes a drastic degradation in model utility during the unlearning process due to the rapid divergence speed of gradient ascent. In this paper, we introduce INPO, an Influence-aware Negative Preference Optimization framework that focuses on slowing the divergence speed and improving the robustness of the model utility to the unlearning process. Specifically, we first analyze that NPO has slower divergence speed and theoretically propose that unlearning high-influence edges can reduce impact of unlearning. We design an influence-aware message function to amplify the influence of unlearned edges and mitigate the tight topological coupling between the forget set and the retain set. The influence of each edge is quickly estimated by a removal-based method. Additionally, we propose a topological entropy loss from the perspective of topology to avoid excessive information loss in the local structure during unlearning. Extensive experiments conducted on five real-world datasets demonstrate that INPO-based model achieves state-of-the-art performance on all forget quality metrics while maintaining the model's utility. Codes are available at https://github.com/sh-qiangchen/INPO.

Large Vision-Language Models (LVLMs) have demonstrated remarkable advancements in numerous areas such as multimedia. However, hallucination issues significantly limit their credibility and application potential. Existing mitigation methods typically rely on external tools or the comparison of multi-round inference, which significantly increase inference time. In this paper, we propose SElf-Evolving Distillation (SEED), which identifies hallucinations within the inner knowledge of LVLMs, isolates and purges them, and then distills the purified knowledge back into the model, enabling self-evolution. Furthermore, we identified that traditional distillation methods are prone to inducing void spaces in the output space of LVLMs. To address this issue, we propose a Mode-Seeking Evolving approach, which performs distillation to capture the dominant modes of the purified knowledge distribution, thereby avoiding the chaotic results that could emerge from void spaces. Moreover, we introduce a Hallucination Elimination Adapter, which corrects the dark knowledge of the original model by learning purified knowledge. Extensive experiments on multiple benchmarks validate the superiority of our SEED, demonstrating substantial improvements in mitigating hallucinations for representative LVLM models such as LLaVA-1.5 and InternVL2. Remarkably, the F1 score of LLaVA-1.5 on the hallucination evaluation metric POPE-Random improved from 81.3 to 88.3.

Vision-Language-Action (VLA) models have demonstrated strong multi-modal reasoning capabilities, enabling direct action generation from visual perception and language instructions in an end-to-end manner. However, their substantial computational cost poses a challenge for real-time robotic control, where rapid decision-making is essential. This paper introduces VLA-Cache, a training-free inference acceleration method that reduces computational overhead by adaptively caching and reusing static visual tokens across frames. Exploiting the temporal continuity in robotic manipulation, VLA-Cache identifies minimally changed tokens between adjacent frames and reuses their cached key-value representations, thereby circumventing redundant computations. Additionally, to maintain action precision, VLA-Cache selectively re-computes task-relevant tokens that are environmentally sensitive, ensuring the fidelity of critical visual information. To further optimize efficiency, we introduce a layer adaptive token reusing strategy that dynamically adjusts the reuse ratio based on attention concentration across decoder layers, prioritizing critical tokens for recomputation. Extensive experiments on two simulation platforms (LIBERO and SIMPLER) and a real-world robotic system demonstrate that VLA-Cache achieves up to 1.7× speedup in CUDA latency and a 15\% increase in control frequency, with negligible loss on task success rate. The code and videos can be found at our project page: https://vla-cache.github.io.

Deep Neural Networks (DNNs) are vulnerable to adversarial attacks. Existing methods are devoted to developing various robust training strategies or regularizations to update the weights of the neural network. But beyond the weights, the overall structure and information flow in the network are explicitly determined by the neural architecture, which remains unexplored. Thus, this paper aims to improve the adversarial robustness of the network from the architectural perspective. We explore the relationship among adversarial robustness, Lipschitz constant, and architecture parameters and show that an appropriate constraint on architecture parameters could reduce the Lipschitz constant to further improve the robustness. The importance of architecture parameters could vary from operation to operation or connection to connection. We approximate the Lipschitz constant of the entire network through a univariate log-normal distribution, whose mean and variance are related to architecture parameters. The confidence can be fulfilled through formulating a constraint on the distribution parameters based on the cumulative function. Compared with adversarially trained neural architectures searched by various NAS algorithms as well as efficient human-designed models, our algorithm empirically achieves the best performance among all the models under various attacks on different datasets.

The paper introduces DIST, an innovative knowledge distillation method that excels in learning from a superior teacher model. DIST differentiates itself from conventional techniques by adeptly handling the often significant prediction discrepancies between the student and teacher models. It achieves this by focusing on maintaining the relationships between their predictions, implementing a correlation-based loss to explicitly capture the teacher's intrinsic inter-class relations. Moreover, DIST uniquely considers the semantic similarities between different instances and each class at the intra-class level. The method is further enhanced by two significant improvements: (1) A teacher acclimation strategy, which effectively reduces the discrepancy between teacher and student, thereby optimizing the distillation process. (2) An extension of the DIST loss from the logit level to the feature level, a modification that proves especially beneficial for dense prediction tasks. DIST stands out for its simplicity, practicality, and adaptability to various architectures, model sizes, and training strategies. It consistently delivers state-of-the-art results across a range of applications, including image classification, object detection, and semantic segmentation.

Understanding long-form videos requires precise temporal action segmentation. While existing studies typically employ multi-stage models that follow an iterative refinement process, we present a novel framework based on the denoising diffusion model that retains this core iterative principle. Within this framework, the model iteratively produces action predictions starting with random noise, conditioned on the features of the input video. To effectively capture three key characteristics of human actions, namely the position prior, the boundary ambiguity, and the relational dependency, we propose a cohesive masking strategy for the conditioning features. Moreover, a consistency gradient guidance technique is proposed, which maximizes the similarity between outputs with or without the masking, thereby enriching conditional information during the inference process. Extensive experiments are performed on four datasets, i.e., GTEA, 50Salads, Breakfast, and Assembly101. The results indicate that our proposed method outperforms or is on par with existing state-of-the-art techniques, underscoring the potential of generative approaches for action segmentation.

This paper defines a positive and unlabeled classification problem for standard GANs, which then leads to a novel technique to stabilize the training of the discriminator in GANs and deal with corrupted data. Traditionally, real data are taken as positive while generated data are negative. This positive-negative classification criterion was kept fixed all through the learning process of the discriminator without considering the gradually improved quality of generated data, even if they could be more realistic than real data at times. In contrast, it is more reasonable to treat the generated data as unlabeled, which could be positive or negative according to their quality. The discriminator is thus a classifier for this positive and unlabeled classification problem, and we derive a new Positive-Unlabeled GAN (PUGAN). We theoretically discuss the global optimality the proposed model will achieve and the equivalent optimization goal. Empirically, we find that PUGAN can achieve comparable or even better performance than those sophisticated discriminator stabilization methods. Considering the potential corrupted data problem in real-world scenarios, we further extend our approach to PUGAN-C, which treats real data as unlabeled that accounts for both clean and corrupted instances, and generated data as positive. The samples from generator could be closer to those corrupted data within unlabeled data at first, but within the framework of adversarial training, the generator will be optimized to cheat the discriminator and produce samples that are similar to those clean data. Experimental results on image generation from several corrupted datasets demonstrate the effectiveness and generalization of PUGAN-C.

Prior research in video object segmentation (VOS) predominantly relies on videos with dense annotations. However, obtaining pixel-level annotations is both costly and time-intensive. In this work, we highlight the potential of effectively training a VOS model using remarkably sparse video annotations-specifically, as few as one or two labeled frames per training video, yet maintaining near equivalent performance levels. We introduce this innovative training methodology as low-shot video object segmentation, abbreviated as low-shot VOS. Central to this method is the generation of reliable pseudo labels for unlabeled frames during the training phase, which are then used in tandem with labeled frames to optimize the model. Notably, our strategy is extremely simple and can be incorporated into the vast majority of current VOS models. For the first time, we propose a universal method for training VOS models on one-shot and two-shot VOS datasets. In the two-shot configuration, utilizing just 7.3% and 2.9% of labeled data from the YouTube-VOS and DAVIS benchmarks respectively, our model delivers results on par with those trained on completely labeled datasets. It is also worth noting that in the one-shot setting, a minor performance decrement is observed in comparison to models trained on fully annotated datasets.

Nowadays, generative models are shaping various fields such as art, design, and human-computer interaction, yet they are accompanied by copyright infringement and content management challenges. In response, existing research seeks to identify the unique fingerprints on the images they generate, which can be leveraged to attribute the generated images to their source models. However, existing methods are restricted to identifying models within a static set included in classifier training, incapable of adapting dynamically to newly emerging unseen models. To bridge this gap, this paper aims to develop a generalized model fingerprint extractor capable of zero-shot attribution that effectively attributes unseen models without exposure during training. Central to our method is a model synthesis technique, which generates numerous synthetic models that mimic the fingerprint patterns of real-world generative models. The design of the synthesis technique is motivated by observations on how the basic generative model's architecture building blocks and parameters influence fingerprint patterns, and it is validated through designed metrics to examine synthetic models' fidelity. Our experiments demonstrate that the fingerprint extractor, trained solely on synthetic models, achieves impressive zero-shot generalization on a wide range of real-world generative models, improving model identification and verification accuracy on unseen models by over 40% and 15%, respectively, compared to existing approaches.

The aim of multi-label few-shot image classification (ML-FSIC) is to assign semantic labels to images, in settings where only a small number of training examples are available for each label. A key feature of the multi-label setting is that an image often has several labels, which typically refer to objects appearing in different regions of the image. When estimating label prototypes, in a metric-based setting, it is thus important to determine which regions are relevant for which labels, but the limited amount of training data and the noisy nature of local features make this highly challenging. As a solution, we propose a strategy in which label prototypes are gradually refined. First, we initialize the prototypes using word embeddings, which allows us to leverage prior knowledge about the meaning of the labels. Second, taking advantage of these initial prototypes, we then use a Loss Change Measurement (LCM) strategy to select the local features from the training images (i.e., the support set) that are most likely to be representative of a given label. Third, we construct the final prototype of the label by aggregating these representative local features using a multi-modal cross-interaction mechanism, which again relies on the initial word embedding-based prototypes. Experiments on COCO, PASCAL VOC, NUS-WIDE, and iMaterialist show that our model substantially improves the current state-of-the-art.

3D style transfer aims to generate stylized views of 3D scenes with specified styles, which requires high-quality generating and keeping multi-view consistency. Existing methods still suffer the challenges of high-quality stylization with texture details and stylization with multimodal guidance. In this paper, we reveal that the common training method of stylization with NeRF, which generates stylized multi-view supervision by 2D style transfer models, causes the same object in supervision to show various states (color tone, details, etc.) in different views, leading NeRF to tend to smooth the texture details, further resulting in low-quality rendering for 3D multi-style transfer. To tackle these problems, we propose a novel Multimodal-guided 3D Multi-style transfer of NeRF, termed MM-NeRF. First, MM-NeRF projects multimodal guidance into a unified space to keep the multimodal styles consistency and extracts multimodal features to guide the 3D stylization. Second, a novel multi-head learning scheme is proposed to relieve the difficulty of learning multi-style transfer, and a multi-view style consistent loss is proposed to track the inconsistency of multi-view supervision data. Finally, a novel incremental learning mechanism is proposed to generalize MM-NeRF to any new style with small costs. Extensive experiments on several real-world datasets show that MM-NeRF achieves high-quality 3D multi-style stylization with multimodal guidance, and keeps multi-view consistency and style consistency between multimodal guidance.

Human motion generation aims to produce plausible human motion sequences according to various conditional inputs, such as text or audio. Despite the feasibility of existing methods in generating motion based on short prompts and simple motion patterns, they encounter difficulties when dealing with long prompts or complex motions. The challenges are two-fold: 1) the scarcity of human motion-captured data for long prompts and complex motions. 2) the high diversity of human motions in the temporal domain and the substantial divergence of distributions from conditional modalities, leading to a many-to-many mapping problem when generating motion with complex and long texts. In this work, we address these gaps by 1) elaborating the first dataset pairing long textual descriptions and 3D complex motions (HumanLong3D), and 2) proposing an autoregressive motion diffusion model (AMD). Specifically, AMD integrates the text prompt at the current timestep with the text prompt and action sequences at the previous timestep as conditional information to predict the current action sequences in an iterative manner. Furthermore, we present its generalization for X-to-Motion with “No Modality Left Behind”, enabling for the first time the generation of high-definition and high-fidelity human motions based on user-defined modality input.

Pretrained diffusion models and their outputs are widely accessible due to their exceptional capacity for synthesizing high-quality images and their open-source nature. The users, however, may face litigation risks owing to the models' tendency to memorize and regurgitate training data during inference. To address this, we introduce Anti-Memorization Guidance (AMG), a novel framework employing three targeted guidance strategies for the main causes of memorization: image and caption duplication, and highly specific user prompts. Consequently, AMG ensures memorization-free outputs while maintaining high Image quality and text alignment, leveraging the synergy of its guidance methods, each indispensable in its own right. AMG also features an innovative automatic detection system for potential memorization during each step of inference process, allows selective application of guidance strategies, minimally interfering with the original sampling process to preserve output utility. We applied AMG to pre-trained Denoising Diffusion Probabilistic Models (DDPM) and Stable Diffusion across various generation tasks. The results demonstrate that AMG is the first approach to successfully eradicates all instances of memorization with no or marginal impacts on image quality and text-alignment, as evidenced by FID and CLIP scores.

Vision Transformers (ViTs) have emerged as a compelling alternative to Convolutional Neural Networks (CNNs) in the realm of computer vision, showcasing tremendous potential. However, recent research has un-veiled a susceptibility of ViTs to adversarial attacks, akin to their CNN counterparts. Adversarial training and randomization are two representative effective defenses for CNNs. Some researchers have attempted to apply adversarial training to ViTs and achieved comparable robustness to CNNs, while it is not easy to directly apply randomization to ViTs because of the architecture difference between CNNs and ViTs. In this paper, we delve into the structural intricacies of ViTs and propose a novel defense mechanism termed Random entangled image Transformer (ReiT), which seamlessly integrates adversarial training and randomization to bolster the adversarial robustness of ViTs. Recognizing the challenge posed by the structural disparities between ViTs and CNNs, we introduce a novel module, input-independent random entangled self-attention (II-ReSA). This module op-timizes random entangled tokens that lead to “dissimilar” self-attention outputs by leveraging model parameters and the sampled random tokens, thereby synthesizing the self-attention module outputs and random entangled tokens to diminish adversarial similarity. ReiT incorporates two distinct random entangled tokens and employs dual randomization, offering an effective countermeasure against adversarial examples while ensuring comprehensive deduction guarantees. Through extensive experiments conducted on various ViT variants and benchmarks, we substantiate the superiority of our proposed method in enhancing the adversarial robustness of Vision Transformers.

Diffusion models (DMs) have achieved remarkable generative performance, particularly with the introduction of stochastic differential equations (SDEs). Nevertheless, a gap emerges in the model sampling trajectory constructed by reverse-SDE due to the accumulation of score estimation and discretization errors. This gap results in a residual in the generated images, adversely impacting the image quality. To remedy this, we propose a novel residual learning framework built upon a correction function. The optimized function enables to improve image quality via rectifying the sampling trajectory effectively. Importantly, our framework exhibits transferable residual correction ability, i.e., a correctionfunction optimized for one pre-trained DM can also enhance the sampling trajectory constructed by other different DMs on the same dataset. Experimental results on four widely-used datasets demonstrate the effectiveness and transferable capability of our framework.

With the transformative impact of the Transformer, DETR pioneered the application of the encoder-decoder ar-chitecture to object detection. A collection of follow-up research, e.g., Deformable DETR, aims to enhance DETR while adhering to the encoder-decoder design. In this work, we revisit the DETR series through the lens of Faster R-CNN. We find that the DETR resonates with the underlying principles of Faster R-CNN's RPN-refiner design but benefits from end-to-end detection owing to the incorpo-ration of Hungarian matching. We systematically adapt the Faster R-CNN towards the Deformable DETR, by in-tegrating or repurposing each component of Deformable DETR, and note that Deformable DETR's improved per-formance over Faster R-CNN is attributed to the adoption of advanced modules such as a superior proposal refiner (e.g., deformable attention rather than RoI Align). When viewing the DETR through the RPN-refiner paradigm, we delve into various proposal refinement techniques such as deformable attention, cross attention, and dynamic convo-lution. These proposal refiners cooperate well with each other; thus, we synergistically combine them to estab-lish a Hybrid Proposal Refiner (HPR). Our HPR is ver-satile and can be incorporated into various DETR de-tectors. For instance, by integrating HPR to a strong DETR detector, we achieve an AP of 54.9 on the COCO benchmark, utilizing a ResNet-50 backbone and a 36-epoch training schedule. Code and models are available at https://github.com/ZhaoJingjing713IHPR.

Large language models (LLMs) offer impressive performance across diverse fields, but their increasing complexity raises both design costs and the need for specialized expertise. These challenges are intensified for Neural Architecture Search (NAS) methods reliant on weight-sharing techniques. This paper introduces GNAS, a new NAS method that boosts the search process with the aid of LLMs for efficient model discovery. With insights from existing architectures, GNAS swiftly identifies superior models that can adapt to changing resource constraints. We provide a mathematical framework to facilitate the transfer of knowledge across different model sizes, thereby improving search efficiency. Our experiments conducted on ImageNet, NAS-Bench-Macro, and ChannelBench-Macro confirm the effectiveness of GNAS across both CNN and Transformer architectures.

Code clone detection aims at finding code fragments with syntactic or semantic similarity. Most of current approaches mainly focus on detecting syntactic similarity while ignoring semantic long-term context alignment, and these detection methods encode the source code using human-designed models, a process which requires both expert input and a significant cost of time for experimentation and refinement. To address these challenges, we introduce the Transformer Code Neural Architecture Search (TCNAS), an approach designed to optimize transformer-based architectures for detection. In TCNAS, all channels are trained and evaluated equitably to enhance search efficiency. Besides, we introduce the dataflow of the code by extracting the semantic information from the code fragments. TCNAS facilitates the discovery of an optimal model structure geared towards the detection, eliminating the need for manual design. The searched optimal architecture is utilized to detect the code pairs. We conduct various empirical experiments on the benchmark, which covering all four types of code clone detection. The results demonstrate our approach consistently yields competitive detection scores across a range of evaluations.

With the increasing number of new neural architecture designs and substantial existing neural architectures, it becomes difficult for the researchers to situate their contributions compared with existing neural architectures or establish the connections between their designs and other relevant ones. To discover similar neural architectures in an efficient and automatic manner, we define a new problem Neural Architecture Retrieval which retrieves a set of existing neural architectures which have similar designs to the query neural architecture. Existing graph pre-training strategies cannot address the computational graph in neural architectures due to the graph size and motifs. To fulfill this potential, we propose to divide the graph into motifs which are used to rebuild the macro graph to tackle these issues, and introduce multi-level contrastive learning to achieve accurate graph representation learning. Extensive evaluations on both human-designed and synthesized neural architectures demonstrate the superiority of our algorithm. Such a dataset which contains 12k real-world network architectures, as well as their embedding, is built for neural architecture retrieval.

Deep neural networks are increasingly utilized in various machine learning tasks. However, as these models grow in complexity, they often face calibration issues, despite enhanced prediction accuracy. Many studies have endeavored to improve calibration performance through the use of specific loss functions, data preprocessing and training frameworks. Yet, investigations into calibration properties have been somewhat overlooked. Our study leverages the Neural Architecture Search (NAS) search space, offering an exhaustive model architecture space for thorough calibration properties exploration. We specifically create a model calibration dataset. This dataset evaluates 90 bin-based and 12 additional calibration measurements across 117,702 unique neural networks within the widely employed NATS-Bench search space. Our analysis aims to answer several longstanding questions in the field, using our proposed dataset: (i) Can model calibration be generalized across different datasets? (ii) Can robustness be used as a calibration measurement? (iii) How reliable are calibration metrics? (iv) Does a post-hoc calibration method affect all models uniformly? (v) How does calibration interact with accuracy? (vi) What is the impact of bin size on calibration measurement? (vii) Which architectural designs are beneficial for calibration? Additionally, our study bridges an existing gap by exploring calibration within NAS. By providing this dataset, we enable further research into NAS calibration. As far as we are aware, our research represents the first large-scale investigation into calibration properties and the premier study of calibration issues within NAS.

Training general-purpose vision models on purely sequential visual data, eschewing linguistic inputs, has heralded a new frontier in visual understanding. These models are intended to not only comprehend but also seamlessly transit to out-of-domain tasks. However, current endeavors are hamstrung by an over-reliance on colossal models, exemplified by models with upwards of 3B parameters, and the necessity for an extensive corpus of visual data, often comprising a staggering 400B tokens. In this paper, we delve into the development of an efficient, autoregression-based vision model, innovatively architected to operate on a limited dataset. We meticulously demonstrate how this model achieves proficiency in a spectrum of visual tasks spanning both high-level and low-level semantic understanding during the testing phase. Our empirical evaluations underscore the model's agility in adapting to various tasks, heralding a significant reduction in the parameter footprint, and a marked decrease in training data requirements, thereby paving the way for more sustainable and accessible advancements in the field of generalist vision models. The code is available at https://github.com/ggjy/DeLVM.

Cross-modal transformers have demonstrated superiority in various vision tasks by effectively integrating different modalities. This paper first critiques prior token exchange methods which replace less informative tokens with inter-modal features, and demonstrate exchange based methods underperform cross-attention mechanisms, while the computational demand of the latter inevitably restricts its use with longer sequences. To surmount the computational challenges, we propose *GeminiFusion*, a pixel-wise fusion approach that capitalizes on aligned cross-modal representations. *GeminiFusion* elegantly combines intra-modal and inter-modal attentions, dynamically integrating complementary information across modalities. We employ a layer-adaptive noise to adaptively control their interplay on a per-layer basis, thereby achieving a harmonized fusion process. Notably, *GeminiFusion* maintains linear complexity with respect to the number of input tokens, ensuring this multimodal framework operates with efficiency comparable to unimodal networks. Comprehensive evaluations across multimodal image-to-image translation, $3$D object detection and arbitrary-modal semantic segmentation tasks, including RGB, depth, LiDAR, event data, etc. demonstrate the superior performance of our *GeminiFusion* against leading-edge techniques. The PyTorch code is available [here](https://github.com/JiaDingCN/GeminiFusion).

Imitation learning has emerged as a promising approach for addressing sequential decision-making problems, with the assumption that expert demonstrations are optimal. However, in real-world scenarios, most demonstrations are often imperfect, leading to challenges in the effectiveness of imitation learning. While existing research has focused on optimizing with imperfect demonstrations, the training typically requires a certain proportion of optimal demonstrations to guarantee performance. To tackle these problems, we propose to purify the potential noises in imperfect demonstrations first, and subsequently conduct imitation learning from these purified demonstrations. Motivated by the success of diffusion model, we introduce a two-step purification via diffusion process. In the first step, we apply a forward diffusion process to smooth potential noises in imperfect demonstrations by introducing additional noise. Subsequently, a reverse generative process is utilized to recover the optimal demonstration from the diffused ones. We provide theoretical evidence supporting our approach, demonstrating that the distance between the purified and optimal demonstration can be bounded. Empirical results on MuJoCo and RoboSuite demonstrate the effectiveness of our method from different aspects.

Diffusion Probabilistic Models (DPMs) show significant potential in image generation, yet their performance hinges on having access to large datasets. Previous works, like Generative Adversarial Networks (GANs), have tackled the limited data problem by transferring pre-trained models learned with sufficient data. However, those methods are hard to be utilized in DPMs since the distinct differences between DPM-based and GAN-based methods, showing in the unique iterative denoising process integral and the need for many timesteps with no-targeted noise in DPMs. In this paper, we propose a novel DPMs-based transfer learning method, ANT, to address the limited data problem. It includes two strategies: similarity-guided training, which boosts transfer with a classifier, and adversarial noise selection which adaptively chooses targeted noise based on the input image. Extensive experiments in the context of few-shot image generation tasks demonstrate that our method is not only efficient but also excels in terms of image quality and diversity when compared to existing GAN-based and DDPM-based methods.

Medical Visual Question Answering (Med-VQA) interprets complex medical imagery using user instructions for precise diagnostics, yet faces challenges due to diverse, inadequately annotated images. In this paper, we introduce the Universal Instruction-Vision Navigator (Uni-Med) framework for extracting instruction-to-answer relationships, facilitating the understanding of visual evidence behind responses. Specifically, we design the Instruct-to-Answer Clues Interpreter (IAI) to generate visual explanations based on the answers and mark the core part of instructions with "real intent" labels. The IAI-Med VQA dataset, produced using IAI, is now publicly available to advance Med-VQA research. Additionally, our Token-Level Cut-Mix module dynamically aligns visual explanations with image patches, ensuring answers are traceable and learnable. We also implement intention-guided attention to minimize non-core instruction interference, sharpening focus on ’real intent’. Extensive experiments on SLAKE datasets show Uni-Med’s superior accuracies (87.52% closed, 86.12% overall), outperforming MedVInT-PMC-VQA by 1.22% and 0.92%. Code and dataset are available at: https://github.com/zhongzee/Uni-Med-master.

Based on data synthesis, anomaly detection (AD) methods often rely on external data for data synthesis. However, most external abnormal data exhibits strong randomness, which may lead to a reduced range of diversity among the synthesized data. In order to achieve a broader diversity in data synthesis, it is necessary to not only have highly diverse data but also to incorporate low-diversity noise data. To enhance the diversity range of the synthesized data, this study proposes a diversity measurement assisted by image self-representation: measuring the distance between noise data and normal data and quantitatively synthesizing diversified data by selecting diverse noise data for synthesis, namely, Diversified Synthesis (DS). Diversified Synthesis introduces patch measurement and a controllable enhancement module to establish controllable diversified enhanced data. The contribution of this study lies in proposing a novel diversified synthesis method, which achieves a broader diversity synthesis through the introduction of image self-representation-assisted diversity measurement and quantitative synthesis. Furthermore, through the self-enhancement data augmentation method, the use of image intrinsic features for enhancement achieves diversity and multi-scale characteristics in the synthesized data, thereby improving the training performance of the discriminative model. This provides an effective optimization solution for comprehensive anomaly detection methods.

Despite the significant achievements of Vision Transformers (ViTs) in various vision tasks, they are constrained by the quadratic complexity. Recently, State Space Models (SSMs) have garnered widespread attention due to their global receptive field and linear complexity with respect to the input length, demonstrating substantial potential across fields including natural language processing and computer vision. To improve the performance of SSMs in vision tasks, a multi-scan strategy is widely adopted, which leads to significant redundancy of SSMs. For a better trade-off between efficiency and performance, we analyze the underlying reasons behind the success of the multi-scan strategy, where long-range dependency plays an important role. Based on the analysis, we introduce Multi-Scale Vision Mamba (MSVMamba) to preserve the superiority of SSMs in vision tasks with limited parameters. It employs a multi-scale 2D scanning technique on both original and downsampled feature maps, which not only benefits long-range dependency learning but also reduces computational costs. Additionally, we integrate a Convolutional Feed-Forward Network (ConvFFN) to address the lack of channel mixing. Our experiments demonstrate that MSVMamba is highly competitive, with the MSVMamba-Tiny model achieving 83.0% top-1 accuracy on ImageNet, 46.9% box mAP, and 42.5% instance mAP with the Mask R-CNN framework, 1x training schedule on COCO, and 47.9% mIoU with single-scale testing on ADE20K. Code is available at https://github.com/YuHengsss/MSVMamba.

Prior research in human-centric AI has primarily addressed single-modality tasks like pedestrian detection, action recognition, and pose estimation. However, the emergence of large multimodal models (LMMs) such as GPT-4V has redirected attention towards integrating language with visual content. Referring expression comprehension (REC) represents a prime example of this multimodal approach. Current human-centric REC benchmarks, typically sourced from general datasets, fall short in the LMM era due to their limitations, such as insufficient testing samples, overly concise referring expressions, and limited vocabulary, making them inadequate for evaluating the full capabilities of modern REC models. In response, we present HC-RefLoCo (Human-Centric Referring Expression Comprehension with Long Context), a benchmark that includes 13,452 images, 24,129 instances, and 44,738 detailed annotations, encompassing a vocabulary of 18,681 words. Each annotation, meticulously reviewed for accuracy, averages 93.2 words and includes topics such as appearance, human-object interaction, location, action, celebrity, and OCR. HC-RefLoCo provides a wider range of instance scales and diverse evaluation protocols, encompassing accuracy with various IoU criteria, scale-aware evaluation, and subject-specific assessments. Our experiments, which assess 24 models, highlight HC-RefLoCo’s potential to advance human-centric AI by challenging contemporary REC models with comprehensive and varied data. Our benchmark, along with the evaluation code, are available at https://github.com/ZhaoJingjing713/HC-RefLoCo.

Tokenizers serve as crucial interfaces between models and linguistic data, substantially influencing the efficacy and precision of large language models (LLMs). Traditional tokenization methods often rely on static frequency-based statistics and are not inherently synchronized with LLM architectures, which may limit model performance. In this study, we propose a simple but effective method to learn tokenizers specifically engineered for seamless integration with LLMs. Initiating with a broad initial vocabulary, we refine our tokenizer by monitoring changes in the model’s perplexity during training, allowing for the selection of a tokenizer that is closely aligned with the model’s evolving dynamics. Through iterative refinement, we develop an optimized tokenizer. Our empirical evaluations demonstrate that this adaptive approach significantly enhances accuracy compared to conventional methods, maintaining comparable vocabulary sizes and affirming its potential to improve LLM functionality.

Self-supervised Learning (SSL) including the mainstream contrastive learning has achieved great success in learning visual representations without data annotations. However, most methods mainly focus on the instance level information (i.e., the different augmented images of the same instance should have the same feature or cluster into the same class), but there is a lack of attention on the relationships between different instances. In this paper, we introduce a novel SSL paradigm, which we term as relational self-supervised learning (ReSSL) framework that learns representations by modeling the relationship between different instances. Specifically, our proposed method employs sharpened distribution of pairwise similarities among different instances as relation metric, which is thus utilized to match the feature embeddings of different augmentations. To boost the performance, we argue that weak augmentations matter to represent a more reliable relation, and leverage momentum strategy for practical efficiency. The designed asymmetric predictor head and an InfoNCE warm-up strategy enhance the robustness to hyper-parameters and benefit the resulting performance. Experimental results show that our proposed ReSSL substantially outperforms the state-of-the-art methods across different network architectures, including various lightweight networks (e.g., EfficientNet and MobileNet).

Authentication and cryptography are critical security functions and, thus, are very often included as part of code. These functions require using credentials, such as passwords, security tokens, and cryptographic keys. However, developers often incorrectly implement/use credentials in their code because of a lack of secure coding skills. This paper analyzes open-source projects concerning the correct use of security credentials. We developed a semantic-rich, language-independent analysis approach for analyzing many projects automatically. We implemented a detection tool, SEAGULL, to automatically check open-source projects based on string literal and code structure information. Instead of analyzing the entire project code, which might result in path explosion when constructing data and control dependencies, SEAGULL pinpoints all literal constants to identify credential candidates and then analyzes the code snippets correlated to these candidates. SEAGULL accurately identifies the leaked credentials by obtaining semantic and syntax information about the code. We applied SEAGULL to 377 open-source projects. SEAGULL successfully reported 19 real-world credential leakages out of those projects. Our analysis shows that some developers protected or erased the credentials in the current project versions, but previously used credentials can still be extracted from the project’s historical versions. Although the implementations of credential leakages seem to be fixed in the current projects, attackers could successfully log into accounts if developers keep using the same credentials as before. Additionally, we found that such credential leakages still affect some projects. By exploiting leaked credentials, attackers can log into particular accounts.

Deep neural networks (DNNs) have achieved state-of-the-art performance on face recognition (FR) tasks in the last decade. In real scenarios, the deployment of DNNs requires taking various face accessories into consideration, like glasses, hats, and masks. In the COVID-19 pandemic era, wearing face masks is one of the most effective ways to defend against the novel coronavirus. However, DNNs are known to be vulnerable to adversarial examples with a small but elaborated perturbation. Thus, a facial mask with adversarial perturbations may pose a great threat to the widely used deep learning-based FR models. In this paper, we consider a challenging adversarial setting: targeted attack against FR models. We propose a new stealthy physical masked FR attack via adversarial style optimization. Specifically, we train an adversarial style mask generator that hides adversarial perturbations inside style masks. Moreover, to ameliorate the phenomenon of sub-optimization with one fixed style, we propose to discover the optimal style given a target through style optimization in a continuous relaxation manner. We simultaneously optimize the generator and the style selection for generating strong and stealthy adversarial style masks. We evaluated the effectiveness and transferability of our proposed method via extensive white-box and black-box digital experiments. Furthermore, we also conducted physical attack experiments against local FR models and online platforms.

Forgery of facial images and videos has increased the concern about digital security. It has led to the significant development of detecting forgery data recently. However, the data, especially the videos published on the Internet, are usually compressed with lossy compression algorithms such as H.264. The compressed data could significantly degrade the performance of recent detection algorithms. The existing anti-compression algorithms focus on enhancing the performance in detecting heavily compressed data but less consider the compression adaption to the data from various compression levels. We believe creating a forgery detection capable of handling data compressed with unknown levels is important. To enhance the performance of such models, we consider the weak compressed and strong compressed data as two views of the original data and they should have similar representation and relationships with other samples. We propose a novel anti-compression forgery detection framework by maintaining closer relations within data under different compression levels. Specifically, our algorithm measures the pair-wise similarity within data as the relations, ensuring that relationships between weakly and strongly compressed data remain consistent. This enhances the discriminative power for detecting highly compressed data. To achieve a better strong compressed data relation guided by the less compressed one, we apply video-level contrastive learning for weak compressed data, which forces the model to produce similar representations within the same video and far from the negative samples. The experiment results show that the proposed algorithm could boost performance for strong compressed data while improving the accuracy rate when detecting clean data.

Recent advanced image restoration (IR) methods typically stack homogeneous operators hierarchically in the UNet architecture. To achieve higher accuracy, these models are now going deeper and more complex, making them resource-intensive. After comprehensively reviewing different operators within modern networks, we provide an in-depth analysis of their individual favorable properties and invent a novel efficient IR network by redesigning the UNet architecture (RUN) with heterogeneous operators. Specifically, we propose three heterogeneous operators for different relational interactions concerning the specificity of different hierarchical features of the UNet architecture. First, the spatial self-attention block (SSA Block) processes high-resolution top-level features by modeling pixel interactions from the spatial dimension. Second, the channel self-attention block (CSA Block) performs channel recalibration and information transmission for the bottom-level features with rich channels. Finally, a simple and efficient convolution block (Conv Block) is used to facilitate middle-order information propagation, which complements the self-attention mechanism to achieve local-global coupling. Based on these designs, our RUN enables more comprehensive information dissemination and interaction regardless of topological distance, thus achieving superior performance while maintaining desirable computational budgets. Extensive experiments show that our RUN achieves state-of-the-art results for a variety of IR tasks, including image deblurring, image denoising, image deraining, and low-light image enhancement.

One major limitation of CNNs is that they are vulnerable to adversarial attacks. Currently, adversarial robustness in neural networks is commonly optimized with respect to a small pre-selected adversarial noise strength, causing them to have potentially limited performance when under attack by larger adversarial noises in real-world scenarios. In this research, we aim to find Neural Architectures that have improved robustness on a wide range of adversarial noise strengths through Neural Architecture Search. In detail, we propose a lightweight Adversarial Noise Estimator to reduce the high cost of generating adversarial noise with respect to different strengths. Besides, we construct an Efficient Wide Spectrum Searcher to reduce the cost of adjusting network architecture with the large adversarial validation set during the search. With the two components proposed, the number of adversarial noise strengths searched can be increased significantly while having a limited increase in search time. Extensive experiments on benchmark datasets such as CIFAR and ImageNet demonstrate that with a significantly richer search signal in robustness, our method can find architectures with improved overall robustness while having a limited impact on natural accuracy and around 40% reduction in search time compared with the naive approach of searching. Codes available at: https://github.com/zhicheng2T0/Wsr-NAS.git

Adversarial imitation learning has become a widely used imitation learning framework. The discriminator is often trained by taking expert demonstrations and policy trajectories as examples respectively from two categories (positive vs. negative) and the policy is then expected to produce trajectories that are indistinguishable from the expert demonstrations. But in the real world, the collected expert demonstrations are more likely to be imperfect, where only an unknown fraction of the demonstrations are optimal. Instead of treating imperfect expert demonstrations as absolutely positive or negative, we investigate unlabeled imperfect expert demonstrations as they are. A positive-unlabeled adversarial imitation learning algorithm is developed to dynamically sample expert demonstrations that can well match the trajectories from the constantly optimized agent policy. The trajectories of an initial agent policy could be closer to those non-optimal expert demonstrations, but within the framework of adversarial imitation learning, agent policy will be optimized to cheat the discriminator and produce trajectories that are similar to those optimal expert demonstrations. Theoretical analysis shows that our method learns from the imperfect demonstrations via a self-paced way. Experimental results on MuJoCo and RoboSuite platforms demonstrate the effectiveness of our method from different aspects.

Recent breakthroughs in semi-supervised semantic segmentation have been developed through contrastive learning. In prevalent pixel-wise contrastive learning solutions, the model maps pixels to deterministic representations and regularizes them in the latent space. However, there exist inaccurate pseudo-labels which map the ambiguous representations of pixels to the wrong classes due to the limited cognitive ability of the model. In this paper, we define pixel-wise representations from a new perspective of probability theory and propose a Probabilistic Representation Contrastive Learning (PRCL) framework that improves representation quality by taking its probability into consideration. Through modelling the mapping from pixels to representations as the probability via multivariate Gaussian distributions, we can tune the contribution of the ambiguous representations to tolerate the risk of inaccurate pseudo-labels. Furthermore, we define prototypes in the form of distributions, which indicates the confidence of a class, while the point prototype cannot. More- over, we propose to regularize the distribution variance to enhance the reliability of representations. Taking advantage of these benefits, high-quality feature representations can be derived in the latent space, thereby the performance of se- mantic segmentation can be further improved. We conduct sufficient experiment to evaluate PRCL on Pascal VOC and CityScapes to demonstrate its superiority. The code is available at https://github.com/Haoyu-Xie/PRCL.

StyleGAN has shown strong potential for disentangled semantic control, thanks to its special design of multi-layer intermediate latent variables. However, existing semantic discovery methods on StyleGAN rely on manual selection of modified latent layers to obtain satisfactory manipulation results, which is tedious and demanding. In this paper, we propose a model that automates this process and achieves state-of-the-art semantic discovery performance. The model consists of an attention-equipped navigator module and losses contrasting deep-feature changes. We propose two model variants, with one contrasting samples in a binary manner, and another one contrasting samples with learned prototype variation patterns. The proposed losses are computed with pretrained deep features, based on our assumption that the features implicitly possess the desired semantic variation structure including consistency and orthogonality. Additionally, we design two metrics to quantitatively evaluate the performance of semantic discovery methods on FFHQ dataset, and also show that disentangled representations can be derived via a simple training process. Experimentally, we show that our models achieve state-of-the-art semantic discovery results without relying on layer-wise manual selection, and these discovered semantics can be used to manipulate real-world images.

Privacy-preserving image generation has been important for segments such as medical domains that have sensitive and limited data. The benefits of guaranteed privacy come at the costs of generated images' quality and utility due to the privacy budget constraints. The utility is currently measured by the gen2real accuracy (g2r%), i.e., the accuracy on real data of a downstream classifier trained using generated data. However, apart from this standard utility, we identify the “reversed utility” as another crucial aspect, which computes the accuracy on generated data of a classifier trained using real data, dubbed as real2gen accuracy (r2g%). Jointly considering these two views of utility, the standard and the reversed, could help the generation model better improve transferability between fake and real data. Therefore, we propose a novel private image generation method that incorporates a dual-purpose auxiliary classifier, which alternates between learning from real data and fake data, into the training of differentially private GANs. Additionally, our deliberate training strategies such as sequential training contributes to accelerating the generator's convergence and further boosting the performance upon exhausting the privacy budget. Our results achieve new state-of-the-arts over all metrics on three benchmarks: MNIST, Fashion-MNIST, and CelebA.

Deep Neural Networks show superior performance in various tasks but are vulnerable to adversarial attacks. Most defense techniques are devoted to the adversarial training strategies, however, it is difficult to achieve satisfactory robust performance only with traditional adversarial training. We mainly attribute it to that aggressive perturbations which lead to the loss increment can always be found via gradient ascent in white-box setting. Although some noises can be involved to prevent attacks from deriving precise gradients on inputs, there exist trade-offs between the defense capability and natural generalization. Taking advantage of the properties of random projection, we propose to replace part of convolutional filters with random projection filters, and theoretically explore the geometric representation preservation of proposed synthesized filters via Johnson-Lindenstrauss lemma. We conduct sufficient evaluation on multiple networks and datasets. The experimental results showcase the superiority of proposed random projection filters to state-of-the-art baselines. The code is available on GitHub.

Although deep neural networks (DNNs) have shown great successes in computer vision tasks, they are vulnerable to perturbations on inputs, and there exists a trade-off between the natural accuracy and robustness to such perturbations, which is mainly caused by the existence of robust non-predictive features and non-robust predictive features. Recent empirical analyses find Vision Transformers (ViTs) are inherently robust to various kinds of perturbations, but the aforementioned trade-off still exists for them. In this work, we propose Trade-off between Robustness and Accuracy of Vision Transformers (TORA-ViTs), which aims to efficiently transfer ViT models pretrained on natural tasks for both accuracy and robustness. TORA-ViTs consist of two major components, including a pair of accuracy and robustness adapters to extract predictive and robust features, respectively, and a gated fusion module to adjust the trade-off. The gated fusion module takes outputs of a pre-trained ViT block as queries and outputs of our adapters as keys and values, and tokens from different adapters at different spatial locations are compared with each other to generate attention scores for a balanced mixing of predictive and robust features. Experiments on ImageNet with various robust benchmarks show that our TORA-ViTs can efficiently improve the robustness of naturally pretrained ViTs while maintaining competitive natural accuracy. Our most balanced setting (TORA-ViTs with λ= 0.5) can maintain 83.7% accuracy on clean ImageNet and reach 54.7% and 38.0% accuracy under FGSM and PGD white-box attacks, respectively. In terms of various ImageNet variants, it can reach 39.2% and 56.3% accuracy on ImageNet-A and ImageNet-R and reach 34.4% mCE on ImageNet-C.

Existing methods of multi-person video 3D human Pose and Shape Estimation (PSE) typically adopt a two-stage strategy, which first detects human instances in each frame and then performs single-person PSE with temporal model. However, the global spatio-temporal context among spatial instances can not be captured. In this paper, we propose a new end-to-end multi-person 3D Pose and Shape estimation framework with progressive Video Transformer, termed PSVT. In PSVT, a spatio-temporal encoder (STE) captures the global feature dependencies among spatial objects. Then, spatio-temporal pose decoder (STPD) and shape decoder (STSD) capture the global dependencies between pose queries and feature tokens, shape queries and feature tokens, respectively. To handle the variances of objects as time proceeds, a novel scheme of progressive decoding is used to update pose and shape queries at each frame. Besides, we propose a novel pose-guided attention (PGA) for shape decoder to better predict shape parameters. The two components strengthen the decoder of PSVT to improve performance. Extensive experiments on the four datasets show that PSVT achieves stage-of-the-art results.

Human-Object Interaction (HOI) detection is a challenging computer vision task that requires visual models to address the complex interactive relationship between humans and objects and predicttriplets. Despite the challenges posed by the numerous interaction combinations, they also offer opportunities for multi-modal learning of visual texts. In this paper, we present a systematic and unified framework (RmLR) that enhances HOI detection by incorporating structured text knowledge. Firstly, we qualitatively and quantitatively analyze the loss of interaction information in the two-stage HOI detector and propose a re-mining strategy to generate more comprehensive visual representation. Secondly, we design more fine-grained sentence- and word-level alignment and knowledge transfer strategies to effectively address the many-to-many matching problem between multiple interactions and multiple texts. These strategies alleviate the matching confusion problem that arises when multiple interactions occur simultaneously, thereby improving the effectiveness of the alignment process. Finally, HOI reasoning by visual features augmented with textual knowledge substantially improves the understanding of interactions. Experimental results illustrate the effectiveness of our approach, where state-of-the-art performance is achieved on public benchmarks.

Recent audio2mesh-based methods have shown promising prospects for speech-driven 3D facial animation tasks. However, some intractable challenges are urgent to be settled. For example, the data-scarcity problem is intrinsically inevitable due to the difficulty of 4D data collection. Besides, current methods generally lack controllability on the animated face. To this end, we propose a novel framework named Speech4Mesh to consecutively generate 4D talking head data and train the audio2mesh network with the reconstructed meshes. In our framework, we first reconstruct the 4D talking head sequence based on the monocular videos. For precise capture of the talking-related variation on the face, we exploit the audio-visual alignment information from the video by employing a contrastive learning scheme. We next can train the audio2mesh network (e.g., FaceFormer) based on the generated 4D data. To get control of the animated talking face, we encode the speaking-unrelated factors (e.g., emotion, etc.) into an emotion embedding for manipulation. Finally, a differentiable renderer guarantees more accurate photometric details of the reconstruction and animation results. Empirical experiments demonstrate that the Speech4Mesh framework can not only outperform state-of-the-art reconstruction methods, especially on the lower-face part but also achieve better animation performance both perceptually and objectively after pre-trained on the synthesized data. Besides, we also verify that the proposed framework is able to explicitly control the emotion of the animated talking face.

Approximately, 350 million people, a proportion of 8%, suffer from color vision deficiency (CVD). While image generation algorithms have been highly successful in synthesizing high-quality images, CVD populations are unintentionally excluded from target users and have difficulties understanding the generated images as normal viewers do. Although a straightforward baseline can be formed by combining generation models and recolor compensation methods as the post-processing, the CVD friendliness of the result images is still limited since the input image content of recolor methods is not CVD-oriented and will be fixed during the recolor compensation process. Besides, the CVD populations can not be fully served since the varying degrees of CVD are often neglected in recoloring methods. Instead, we propose a personalized CVD-friendly image generation algorithm with two key characteristics: (i) generating CVD-oriented images aligned with the needs of CVD populations; (ii) generating continuous personalized images for people with various CVD degrees through disentangling the color representation based on a triple-latent structure. Quantitative and qualitative experiments indicate our proposed image generation model can generate practical and compelling results compared to the normal generation model and combination baselines on several datasets. The code is available at: https://github.com/Jiangshuyi0V0/CVD-GAN.git

Temporal action segmentation is crucial for understanding long-form videos. Previous works on this task commonly adopt an iterative refinement paradigm by using multi-stage models. We propose a novel framework via denoising diffusion models, which nonetheless shares the same inherent spirit of such iterative refinement. In this framework, action predictions are iteratively generated from random noise with input video features as conditions. To enhance the modeling of three striking characteristics of human actions, including the position prior, the boundary ambiguity, and the relational dependency, we devise a unified masking strategy for the conditioning inputs in our framework. Extensive experiments on three benchmark datasets, i.e., GTEA, 50Salads, and Breakfast, are performed and the proposed method achieves superior or comparable results to state-of-the-art methods, showing the effectiveness of a generative approach for action segmentation. Code is at tinyurl.com/DiffAct.

Temporal modeling plays a crucial role in understanding video content. To tackle this problem, previous studies built complicated temporal relations through time sequence thanks to the development of computationally powerful devices. In this work, we explore the potential of four simple arithmetic operations for temporal modeling. Specifically, we first capture auxiliary temporal cues by computing addition, subtraction, multiplication, and division between pairs of extracted frame features. Then, we extract corresponding features from these cues to benefit the original temporal-irrespective domain. We term such a simple pipeline as an Arithmetic Temporal Module (ATM), which operates on the stem of a visual backbone with a plug-and-play style. We conduct comprehensive ablation studies on the instantiation of ATMs and demonstrate that this module provides powerful temporal modeling capability at a low computational cost. Moreover, the ATM is compatible with both CNNs- and ViTs-based architectures. Our results show that ATM achieves superior performance over several popular video benchmarks. Specifically, on Something-Something V1, V2 and Kinetics-400, we reach top-1 accuracy of 65.6%, 74.6%, and 89.4% respectively. The code is available at https://github.com/whwu95/ATM.

Semi-Supervised image classification is one of the most fundamental problem in computer vision, which significantly reduces the need for human labor. In this paper, we introduce a new semi-supervised learning algorithm - Sim-MatchV2, which formulates various consistency regularizations between labeled and unlabeled data from the graph perspective. In SimMatchV2, we regard the augmented view of a sample as a node, which consists of a label and its corresponding representation. Different nodes are connected with the edges, which are measured by the similarity of the node representations. Inspired by the message passing and node classification in graph theory, we propose four types of consistencies, namely 1) node-node consistency, 2) node-edge consistency, 3) edge-edge consistency, and 4) edge-node consistency. We also uncover that a simple feature normalization can reduce the gaps of the feature norm between different augmented views, significantly improving the performance of SimMatchV2. Our SimMatchV2 has been validated on multiple semi-supervised learning benchmarks. Notably, with ResNet-50 as our backbone and 300 epochs of training, SimMatchV2 achieves 71.9% and 76.2% Top-1 Accuracy with 1% and 10% labeled examples on ImageNet, which significantly outperforms the previous methods and achieves state-of-the-art performance.

Diffusion models have recently exhibited remarkable performance on synthetic data. After a diffusion path is selected, a base model, such as UNet, operates as a denoising autoencoder, primarily predicting noises that need to be eliminated step by step. Consequently, it is crucial to employ a model that aligns with the expected budgets to facilitate superior synthetic performance. In this paper, we meticulously analyze the diffusion model and engineer a base model search approach, denoted "DiffNAS". Specifically, we leverage GPT-4 as a supernet to expedite the search, supplemented with a search memory to enhance the results. Moreover, we employ RFID as a proxy to promptly rank the experimental outcomes produced by GPT-4. We also adopt a rapid-convergence training strategy to boost search efficiency. Rigorous experimentation corroborates that our algorithm can augment the search efficiency by $2 \times$ under GPT-based scenarios, while also attaining a performance of 2.82 with 0.37 improvement in FID on CIFAR10 relative to the benchmark IDDPM algorithm.

Diffusion models recently achieved state-of-the-art in image generation. They mainly utilize the denoising framework, which leverages the Langevin dynamics process for image sampling. Recently, the guidance method has modified this process to add conditional information to achieve a controllable generator. However, the current guidance on denoising processes suffers from the trade-off between diversity, image quality, and conditional information. In this work, we propose to view this guidance sampling process from a gradient view, where image pixels are treated as parameters being optimized, and each mathematical term in the sampling process represents one update direction. This perspective reveals more insights into the conflict problems between updated directions on the pixels, which cause the trade-off as mentioned previously. We investigate the conflict problems and propose to solve them by a simple projection method. The experimental results evidently improve over different baselines on datasets with various resolutions.

The use of deep neural networks in real-world applications require well-calibrated networks with confidence scores that accurately reflect the actual probability. However, it has been found that these networks often provide over-confident predictions, which leads to poor calibration. Recent efforts have sought to address this issue by focal loss to reduce over-confidence, but this approach can also lead to under-confident predictions. While different variants of focal loss have been explored, it is difficult to find a balance between over-confidence and under-confidence. In our work, we propose a new loss function by focusing on dual logits. Our method not only considers the ground truth logit, but also take into account the highest logit ranked after the ground truth logit. By maximizing the gap between these two logits, our proposed dual focal loss can achieve a better balance between over-confidence and under-confidence. We provide theoretical evidence to support our approach and demonstrate its effectiveness through evaluations on multiple models and datasets, where it achieves state-of-the-art performance. Code is available at https://github.com/Linwei94/DualFocalLoss

Confidence calibration - the process to calibrate the output probability distribution of neural networks - is essential for safety-critical applications of such networks. Recent works verify the link between mis-calibration and overfitting. However, early stopping, as a well-known technique to mitigate overfitting, fails to calibrate networks. In this work, we study the limitions of early stopping and comprehensively analyze the overfitting problem of a network considering each individual block. We then propose a novel regularization method, predecessor combination search (PCS), to improve calibration by searching a combination of best-fitting block predecessors, where block predecessors are the corresponding network blocks with weight parameters from earlier training stages. PCS achieves the state-of-the-art calibration performance on multiple datasets and architectures. In addition, PCS improves model robustness under dataset distribution shift. Supplementary material and code are available at https://github.com/Linwei94/PCS

Multitask learning combines both commons among a group of tasks and each task's individual characters to achieve better performance. However, different tasks may lead to contrary gradient directions or imbalance gradient values disrupting the training process. We propose a new regularization algorithm based on multitask local Rademacher complexity called MultiDrop, dropping some random tasks in optimization. A new regularization function has been derived based on the upper bound of local Rademacher complexity in multitask learning. Experiments on different image classification datasets have been implemented to demonstrate the effectiveness of MultiDrop compared with other regularization algorithms, and links between theoretical analyses and experiments.

Recently, the field of generative models has seen a significant advancement with the introduction of Diffusion Probabilistic Models (DPMs). The Denoising Diffusion Implicit Model (DDIM) was designed to reduce computational time by skipping a number of steps in the inference process of DPMs. However, the hand-crafted sampling schedule in DDIM, which relies on human expertise, has its limitations in considering all relevant factors in the sampling process. Additionally, the assumption that all instances should have the same schedule is not always valid. To address these problems, this paper proposes a method that leverages reinforcement learning to automatically search for an optimal sampling schedule for DPMs. This is achieved by a policy network that predicts the next step to visit based on the current state of the noisy image. The optimization of the policy network is accomplished using an episodic actor-critic framework, which incorporates reinforcement learning. Empirical results demonstrate the superiority of our approach over various datasets with different timesteps. We also observe that the trained sampling schedule has a strong generalization ability across different DPM baselines.

Human-object interaction (HOI) detection aims to comprehend the intricate relationships between humans and objects, predicting <human, action, object> triplets, and serving as the foundation for numerous computer vision tasks. The complexity and diversity of human-object interactions in the real world, however, pose significant challenges for both annotation and recognition, particularly in recognizing interactions within an open world context. This study explores the universal interaction recognition in an open-world setting through the use of Vision-Language (VL) foundation models and large language models (LLMs). The proposed method is dubbed as UniHOI. We conduct a deep analysis of the three hierarchical features inherent in visual HOI detectors and propose a method for high-level relation extraction aimed at VL foundation models, which we call HO prompt-based learning. Our design includes an HO Prompt-guided Decoder (HOPD), facilitates the association of high-level relation representations in the foundation model with various HO pairs within the image. Furthermore, we utilize a LLM (i.e. GPT) for interaction interpretation, generating a richer linguistic understanding for complex HOIs. For open-category interaction recognition, our method supports either of two input types: interaction phrase or interpretive sentence. Our efficient architecture design and learning methods effectively unleash the potential of the VL foundation models and LLMs, allowing UniHOI to surpass all existing methods with a substantial margin, under both supervised and zero-shot settings. The code and pre-trained weights will be made publicly available.

This paper tackles two critical challenges encountered in classifier guidance for diffusion generative models, i.e., the lack of diversity and the presence of adversarial effects. These issues often result in a scarcity of diverse samples or the generation of non-robust features. The underlying cause lies in the mechanism of classifier guidance, where discriminative gradients push samples to be recognized as conditions aggressively. This inadvertently suppresses information with common features among relevant classes, resulting in a limited pool of features with less diversity or the absence of robust features for image construction. We propose a generalized classifier guidance method called Progressive Guidance, which mitigates the problems by allowing relevant classes' gradients to contribute to shared information construction when the image is noisy in early sampling steps. In the later sampling stage, we progressively enhance gradients to refine the details in the image toward the primary condition. This helps to attain a high level of diversity and robustness compared to the vanilla classifier guidance. Experimental results demonstrate that our proposed method further improves the image quality while offering a significant level of diversity as well as robust features.

Recently, text-to-image models have been thriving. Despite their powerful generative capacity, our research has uncovered a lack of robustness in this generation process. Specifically, the introduction of small perturbations to the text prompts can result in the blending of primary subjects with other categories or their complete disappearance in the generated images. In this paper, we propose **Auto-attack on Text-to-image Models (ATM)**, a gradient-based approach, to effectively and efficiently generate such perturbations. By learning a Gumbel Softmax distribution, we can make the discrete process of word replacement or extension continuous, thus ensuring the differentiability of the perturbation generation. Once the distribution is learned, ATM can sample multiple attack samples simultaneously. These attack samples can prevent the generative model from generating the desired subjects without tampering with the category keywords in the prompt. ATM has achieved a 91.1\% success rate in short-text attacks and an 81.2\% success rate in long-text attacks. Further empirical analysis revealed three attack patterns based on: 1) variability in generation speed, 2) similarity of coarse-grained characteristics, and 3) polysemy of words. The code is available at https://github.com/duchengbin8/Stable_Diffusion_is_Unstable

The tremendous success of large models trained on extensive datasets demonstrates that scale is a key ingredient in achieving superior results. Therefore, the reflection on the rationality of designing knowledge distillation (KD) approaches for limited-capacity architectures solely based on small-scale datasets is now deemed imperative. In this paper, we identify the small data pitfall that presents in previous KD methods, which results in the underestimation of the power of vanilla KD framework on large-scale datasets such as ImageNet-1K. Specifically, we show that employing stronger data augmentation techniques and using larger datasets can directly decrease the gap between vanilla KD and other meticulously designed KD variants. This highlights the necessity of designing and evaluating KD approaches in the context of practical scenarios, casting off the limitations of small-scale datasets. Our investigation of the vanilla KD and its variants in more complex schemes, including stronger training strategies and different model capacities, demonstrates that vanilla KD is elegantly simple but astonishingly effective in large-scale scenarios. Without bells and whistles, we obtain state-of-the-art ResNet-50, ViT-S, and ConvNeXtV2-T models for ImageNet, which achieve 83.1%, 84.3%, and 85.0% top-1 accuracy, respectively. PyTorch code and checkpoints can be found at https://github.com/Hao840/vanillaKD.

Knowledge distillation (KD) has proven to be a highly effective approach for enhancing model performance through a teacher-student training scheme. However, most existing distillation methods are designed under the assumption that the teacher and student models belong to the same model family, particularly the hint-based approaches. By using centered kernel alignment (CKA) to compare the learned features between heterogeneous teacher and student models, we observe significant feature divergence. This divergence illustrates the ineffectiveness of previous hint-based methods in cross-architecture distillation. To tackle the challenge in distilling heterogeneous models, we propose a simple yet effective one-for-all KD framework called OFA-KD, which significantly improves the distillation performance between heterogeneous architectures. Specifically, we project intermediate features into an aligned latent space such as the logits space, where architecture-specific information is discarded. Additionally, we introduce an adaptive target enhancement scheme to prevent the student from being disturbed by irrelevant information. Extensive experiments with various architectures, including CNN, Transformer, and MLP, demonstrate the superiority of our OFA-KD framework in enabling distillation between heterogeneous architectures. Specifically, when equipped with our OFA-KD, the student models achieve notable performance improvements, with a maximum gain of 8.0% on the CIFAR-100 dataset and 0.7% on the ImageNet-1K dataset. PyTorch code and checkpoints can be found at https://github.com/Hao840/OFAKD.

The representation gap between teacher and student is an emerging topic in knowledge distillation (KD). To reduce the gap and improve the performance, current methods often resort to complicated training schemes, loss functions, and feature alignments, which are task-specific and feature-specific. In this paper, we state that the essence of these methods is to discard the noisy information and distill the valuable information in the feature, and propose a novel KD method dubbed DiffKD, to explicitly denoise and match features using diffusion models. Our approach is based on the observation that student features typically contain more noises than teacher features due to the smaller capacity of student model. To address this, we propose to denoise student features using a diffusion model trained by teacher features. This allows us to perform better distillation between the refined clean feature and teacher feature. Additionally, we introduce a light-weight diffusion model with a linear autoencoder to reduce the computation cost and an adaptive noise matching module to improve the denoising performance. Extensive experiments demonstrate that DiffKD is effective across various types of features and achieves state-of-the-art performance consistently on image classification, object detection, and semantic segmentation tasks. Code is available at https://github.com/hunto/DiffKD.

Deep neural networks have been found to be vulnerable in a variety of tasks. Adversarial attacks can manipulate network outputs, resulting in incorrect predictions. Adversarial defense methods aim to improve the adversarial robustness of networks by countering potential attacks. In addition to traditional defense approaches, randomized defense mechanisms have recently received increasing attention from researchers. These methods introduce different types of perturbations during the inference phase to destabilize adversarial attacks. Although promising empirical results have been demonstrated by these approaches, the defense performance is quite sensitive to the randomness parameters, which are always manually tuned without further analysis. On the contrary, we propose incorporating random weights into the optimization to fully exploit the potential of randomized defense. To perform better optimization of randomness parameters, we conduct a theoretical analysis of the connections between randomness parameters and gradient similarity as well as natural performance. From these two aspects, we suggest imposing theoretically-guided constraints on random weights during optimizations, as these weights play a critical role in balancing natural performance and adversarial robustness. We derive both the upper and lower bounds of random weight parameters by considering prediction bias and gradient similarity. In this study, we introduce the Constrained Trainable Random Weight (CTRW), which adds random weight parameters to the optimization and includes a constraint guided by the upper and lower bounds to achieve better trade-offs between natural and robust accuracy. We evaluate the effectiveness of CTRW on several datasets and benchmark convolutional neural networks. Our results indicate that our model achieves a robust accuracy approximately 16% to 17% higher than the baseline model under PGD-20 and 22% to 25% higher on Auto Attack.

Recent advancements in masked image modeling (MIM) have made it a prevailing framework for self-supervised visual representation learning. The MIM pretrained models, like most deep neural network methods, remain vulnerable to adversarial attacks, limiting their practical application, and this issue has received little research attention. In this paper, we investigate how this powerful self-supervised learning paradigm can provide adversarial robustness to downstream classifiers. During the exploration, we find that noisy image modeling (NIM), a simple variant of MIM that adopts denoising as the pre-text task, reconstructs noisy images surprisingly well despite severe corruption. Motivated by this observation, we propose an adversarial defense method, referred to as De^3, by exploiting the pretrained decoder for denoising. Through De^3, NIM is able to enhance adversarial robustness beyond providing pretrained features. Furthermore, we incorporate a simple modification, sampling the noise scale hyperparameter from random distributions, and enable the defense to achieve a better and tunable trade-off between accuracy and robustness. Experimental results demonstrate that, in terms of adversarial robustness, NIM is superior to MIM thanks to its effective denoising capability. Moreover, the defense provided by NIM achieves performance on par with adversarial training while offering the extra tunability advantage. Source code and models are available at https://github.com/youzunzhi/NIM-AdvDef.

The past few years have witnessed great success in the use of diffusion models (DMs) to generate high-fidelity images with the help of stochastic differential equations (SDEs). However, discretization error is an inevitable limitation when utilizing numerical solvers to solve SDEs. To address this limitation, we provide a theoretical analysis demonstrating that an appropriate combination of the contrastive loss and score matching serves as an upper bound of the KL divergence between the true data distribution and the model distribution. To obtain this bound, we utilize a contrastive loss to construct a contrastive sampling chain to fine-tuning the pre-trained DM. In this manner, our method reduces the discretization error and thus yields a smaller gap between the true data distribution and our model distribution. Moreover, the presented method can be applied to fine-tuning various pre-trained DMs, both with or without fast sampling algorithms, contributing to better sample quality or slightly faster sampling speeds. To validate the efficacy of our method, we conduct comprehensive experiments. For example, on CIFAR10, when applied to a pre-trained EDM, our method improves the FID from 2.04 to 1.88 with 35 neural function evaluations (NFEs), and reduces NFEs from 35 to 25 to achieve the same 2.04 FID.

This paper searches for the optimal neural architecture by minimizing a proxy of validation loss. Existing neural architecture search (NAS) methods used to discover the optimal neural architecture that best fits the validation examples given the up-to-date network weights. These intermediate validation results are invaluable but have not been fully explored. We propose to approximate the validation loss landscape by learning a mapping from neural architectures to their corresponding validate losses. The optimal neural architecture thus can be easily identified as the minimum of this proxy validation loss landscape. To improve the efficiency, a novel architecture sampling strategy is developed for the approximation of the proxy validation loss landscape. We also propose an operation importance weight (OIW) to balance the randomness and certainty of architecture sampling. The representation of neural architecture is learned through a graph autoencoder (GAE) over both architectures sampled during search and randomly generated architectures. We provide theoretical analyses on the validation loss estimator learned with our sampling strategy. Experimental results demonstrate that the proposed proxy validation loss landscape can be effective in both the differentiable NAS and the evolutionary-algorithm-based (EA-based) NAS.

Video Super-Resolution (VSR) aims to restore high-resolution (HR) videos from low-resolution (LR) videos. Existing VSR techniques usually recover HR frames by extracting pertinent textures from nearby frames with known degradation processes. Despite significant progress, grand challenges remain to effectively extract and transmit high-quality textures from high-degraded low-quality sequences, such as blur, additive noises, and compression artifacts. This work proposes a novel degradation-robust Frequency-Transformer (FTVSR++) for handling low-quality videos that carry out self-attention in a combined space-time-frequency domain. First, video frames are split into patches and each patch is transformed into spectral maps in which each channel represents a frequency band. It permits a fine-grained self-attention on each frequency band so that real visual texture can be distinguished from artifacts. Second, a novel dual frequency attention (DFA) mechanism is proposed to capture the global and local frequency relations, which can handle different complicated degradation processes in real-world scenarios. Third, we explore different self-attention schemes for video processing in the frequency domain and discover that a "divided attention" which conducts joint space-frequency attention before applying temporal-frequency attention, leads to the best video enhancement quality. Extensive experiments on three widely-used VSR datasets show that FTVSR++ outperforms state-of-the-art methods on different low-quality videos with clear visual margins.

Searching for a more compact network width recently serves as an effective way of channel pruning for the deployment of convolutional neural networks (CNNs) under hardware constraints. To fulfil the searching, a one-shot supernet is usually leveraged to efficiently evaluate the performance w.r.t. different network widths. However, current methods mainly follow a unilaterally augmented (UA) principle for the evaluation of each width, which induces the training unfairness of channels in supernet. In this article, we introduce a new supernet called Bilaterally Coupled Network (BCNet) to address this issue. In BCNet, each channel is fairly trained and responsible for the same amount of network widths, thus each network width can be evaluated more accurately. Besides, we propose to reduce the redundant search space and present the BCNetV2 as the enhanced supernet to ensure rigorous training fairness over channels. Furthermore, we leverage a stochastic complementary strategy for training the BCNet, and propose a prior initial population sampling method to boost the performance of the evolutionary search. We also propose a new open-source width search benchmark on macro structures named Channel-Bench-Macro for the better comparisons of the width search algorithms with MobileNet- and ResNet-like architectures. Extensive experiments on the benchmark datasets demonstrate that our method can achieve state-of-the-art performance.

Deep convolutional neural networks (CNNs) have demonstrated tremendous success in image super-resolution (SR). According to the frequency principle, the vanilla CNNs fit the target function from low to high frequencies during the training process. It implies an implicit bias that CNNs tend to fit the training data by a low-frequency function. This is detrimental to SR task which essentially recovers the missing high-frequency cues from degraded images. To address this issue, a novel spatial-frequency domain hybrid network (SFHN) is proposed for image SR in this paper. More specifically, it contains multiple spatial-frequency domain hybrid convolution blocks (SFBlocks) for extracting both spatial and frequency components. In particular, frequency information is complementary to spatial one, mitigating the disadvantage of the spatial CNNs in capturing high-frequency content resulting from the inherent bias of the networks. In contrast to convolution blocks in the previous SR methods, the proposed SFBlock is configured with additional frequency-domain convolution branch driven by the Fourier transform, which allows to process frequency content. In addition, we introduce a spectral loss based on Fast Fourier Transform (FFT) to fully leverage the performance of our SFHN, as it prevents the loss of important frequency content during training. Experimental results on public benchmarks demonstrate that our SFHN achieves promising performance superior to the state-of-the-art methods.

To achieve efficient inference with a hardware-friendly design, Adder Neural Networks (ANNs) are proposed to replace expensive multiplication operations in Convolutional Neural Networks (CNNs) with cheap additions through utilizing ℓ1-norm for similarity measurement instead of cosine distance. However, we observe that there exists an increasing gap between CNNs and ANNs with reducing parameters, which cannot be eliminated by existing algorithms. In this paper, we present a simple yet effective Norm-Guided Distillation (NGD) method for ℓ1-norm ANNs to learn superior performance from ℓ2-norm ANNs. Although CNNs achieve similar accuracy with ℓ2-norm ANNs, the clustering performance based on ℓ2-distance can be easily learned by ℓ1-norm ANNs compared with cross correlation in CNNs. The features in ℓ2-norm ANNs are encouraged to achieve intra-class centralization and inter-class decentralization to amplify this advantage. Furthermore, the roughly estimated gradients in vanilla ANNs are modified to a progressive approximation from ℓ2-norm to ℓ1-norm so that a more accurate optimization can be achieved. Extensive evaluations on several benchmarks demonstrate the effectiveness of NGD on lightweight networks. For example, our method improves ANN by 10.43% with 0.25x GhostNet on CIFAR-100 and 3.1% with 1.0x GhostNet on ImageNet.

Existing supervised methods have achieved impressive performance in forecasting skeleton-based human motion. However, they often rely on action class labels in both training and inference phases. In practice, it could be a burden to request action class labels in the inference phase, and even for the training phase, the collected labels could be incomplete for sequences with a mixture of multiple actions. In this article, we take action class labels as a kind of privileged supervision that only exists in the training phase. We design a new architecture that includes a motion classification as an auxiliary task with motion prediction. To deal with potential missing labels of motion sequence, we propose a new classification loss function to exploit their relationships with those observed labels and a perceptual loss to measure the difference between ground truth sequence and generated sequence in the classification task. Experimental results on the most challenging Human 3.6M dataset and the Carnegie Mellon University (CMU) dataset demonstrate the effectiveness of the proposed algorithm to exploit action class labels for improved modeling of human dynamics.

This article revisits the bilinear attention networks (BANs) in the visual question answering task from a graph perspective. The classical BANs build a bilinear attention map to extract the joint representation of words in the question and objects in the image but lack fully exploring the relationship between words for complex reasoning. In contrast, we develop bilinear graph networks to model the context of the joint embeddings of words and objects. Two kinds of graphs are investigated, namely, image-graph and question-graph. The image-graph transfers features of the detected objects to their related query words, enabling the output nodes to have both semantic and factual information. The question-graph exchanges information between these output nodes from image-graph to amplify the implicit yet important relationship between objects. These two kinds of graphs cooperate with each other, and thus, our resulting model can build the relationship and dependency between objects, which leads to the realization of multistep reasoning. Experimental results on the VQA v2.0 validation dataset demonstrate the ability of our method to handle complex questions. On the test-std set, our best single model achieves state-of-the-art performance, boosting the overall accuracy to 72.56%, and we are one of the top-two entries in the VQA Challenge 2020.

For the real-world time series analysis, data missing is a ubiquitously existing problem due to anomalies during data collecting and storage. If not treated properly, this problem will seriously hinder the classification, regression, or related tasks. Existing methods for time series imputation either impose too strong assumptions on the distribution of missing data or cannot fully exploit, even simply ignore, the informative temporal dependencies and feature correlations across different time steps. In this article, inspired by the idea of conditional generative adversarial networks, we propose a generative adversarial learning framework for time series imputation under the condition of observed data (as well as the labels, if possible). In our model, we employ a modified bidirectional RNN structure as the generator G, which is aimed at generating the missing values by taking advantage of the temporal and nontemporal information extracted from the observed time series. The discriminator D is designed to distinguish whether each value in a time series is generated or not so that it can help the generator to make an adjustment toward a more authentic imputation result. For an empirical verification of our model, we conduct imputation and classification experiments on several real-world time series data sets. The experimental results show an eminent improvement compared with state-of-the-art baseline models.

Training a good supernet in one-shot NAS methods is difficult since the search space is usually considerably huge$(\mathrm{e}.\mathrm{g}.,\ 13^{21})$. In order to enhance the supernet's evaluation ability, one greedy strategy is to sample good paths, and let the supernet lean towards the good ones and ease its evaluation burden as a result. However, in practice the search can be still quite inefficient since the identification of good paths is not accurate enough and sampled paths still scatter around the whole search space. In this paper, we leverage an explicit path filter to capture the characteristics of paths and directly filter those weak ones, so that the search can be thus implemented on the shrunk space more greedily and efficiently. Concretely, based on the fact that good paths are much less than the weak ones in the space, we argue that the label of “weak paths” will be more confident and reliable than that of “good paths” in multi-path sampling. In this way, we thus cast the training of path filter in the positive and unlabeled (PU) learning paradigm, and also encourage a path embedding as better path/operation representation to enhance the identification capacity of the learned filter. By dint of this embedding, we can further shrink the search space by aggregating similar operations with similar embeddings, and the search can be more efficient and accurate. Extensive experiments validate the effectiveness of the proposed method GreedyNASv2. For example, our obtained GreedyNASv2-L achieves 81.1% Top-1 accuracy on ImageNet dataset, significantly outperforming the ResNet-50 strong baselines.

Structural re-parameterization (Rep) methods achieve noticeable improvements on simple VGG-style networks. Despite the prevalence, current Rep methods simply re-parameterize all operations into an augmented network, including those that rarely contribute to the model's performance. As such, the price to pay is an expensive computational overhead to manipulate these unnecessary behaviors. To eliminate the above caveats, we aim to boot-strap the training with minimal cost by devising a dynamic re-parameterization (DyRep) method, which encodes Rep technique into the training process that dynamically evolves the network structures. Concretely, our proposal adaptively finds the operations which contribute most to the loss in the network, and applies Rep to enhance their representational capacity. Besides, to suppress the noisy and redundant operations introduced by Rep, we devise a de-parameterization technique for a more compact re-parameterization. With this regard, DyRep is more efficient than Rep since it smoothly evolves the given network instead of constructing an over-parameterized network. Experimental results demonstrate our effectiveness, e.g., DyRep improves the accuracy of ResNet-18 by 2.04% on ImageNet and reduces 22% runtime over the baseline. Code is avail-able at: https://github.com/hunto/DyRep.

Vision transformers have been successfully applied to image recognition tasks due to their ability to capture long-range dependencies within an image. However, there are still gaps in both performance and computational cost between transformers and existing convolutional neural networks (CNNs). In this paper, we aim to address this issue and develop a network that can outperform not only the canonical transformers, but also the high-performance convolutional models. We propose a new transformer based hybrid network by taking advantage of transformers to capture long-range dependencies, and of CNNs to extract local information. Furthermore, we scale it to obtain a family of models, called CMTs, obtaining much better trade-off for accuracy and efficiency than previous CNN-based and transformer-based models. In particular, our CMT-S achieves 83.5% top-1 accuracy on ImageNet, while being 14x and 2x smaller on FLOPs than the existing DeiT and EfficientNet, respectively. The proposed CMT-S also generalizes well on CIFAR10 (99.2%), CIFAR100 (91.7%), Flowers (98.7%), and other challenging vision datasets such as COCO (44.3% mAP), with considerably less computational cost.

Previous vision MLPs such as MLP-Mixer and ResMLP accept linearly flattened image patches as input, making them inflexible for different input sizes and hard to capture spatial information. Such approach withholds MLPs from getting comparable performance with their transformer-based counterparts and prevents them from becoming a general backbone for computer vision. This paper presents Hire-MLP, a simple yet competitive vision MLP architecture via Hierarchical rearrangement, which contains two levels of rearrangements. Specifically, the inner-region rearrangement is proposed to capture local information inside a spatial region, and the cross-region rearrangement is proposed to enable information communication between different regions and capture global context by circularly shifting all tokens along spatial directions. Extensive experiments demonstrate the effectiveness of Hire-MLP as a versatile backbone for various vision tasks. In particular, Hire-MLP achieves competitive results on image classification, object detection and semantic segmentation tasks, e.g., 83.8% top-1 accuracy on ImageNet, 51.7% box AP and 44.8% mask AP on COCO val2017, and 49.9% mIoU on ADE20K, surpassing previous transformer-based and MLP-based models with better trade-off for accuracy and throughput.

This paper studies the efficiency problem for visual transformers by excavating redundant calculation in given networks. The recent transformer architecture has demonstrated its effectiveness for achieving excellent performance on a series of computer vision tasks. However, similar to that of convolutional neural networks, the huge computational cost of vision transformers is still a severe issue. Considering that the attention mechanism aggregates different patches layer-by-layer, we present a novel patch slimming approach that discards useless patches in a topdown paradigm. We first identify the effective patches in the last layer and then use them to guide the patch selection process of previous layers. For each layer, the impact of a patch on the final output feature is approximated and patches with less impacts will be removed. Experimental results on benchmark datasets demonstrate that the proposed method can significantly reduce the computational costs of vision transformers without affecting their performances. For example, over 45% FLOPs of the ViT-Ti model can be reduced with only 0.2% top-1 accuracy drop on the ImageNet dataset.

In the field of computer vision, recent works show that a pure MLP architecture mainly stacked by fully-connected layers can achieve competing performance with CNN and transformer. An input image of vision MLP is usually split into multiple tokens (patches), while the existing MLP models directly aggregate them with fixed weights, neglecting the varying semantic information of tokens from different images. To dynamically aggregate tokens, we propose to represent each token as a wave function with two parts, amplitude and phase. Amplitude is the original feature and the phase term is a complex value changing according to the semantic contents of input images. Introducing the phase term can dynamically modulate the relationship between tokens and fixed weights in MLP. Based on the wave-like token representation, we establish a novel Wave-MLP architecture for vision tasks. Extensive experiments demonstrate that the proposed Wave-MLP is superior to the state-of-the-art MLP architectures on various vision tasks such as image classification, object detection and semantic segmentation. The source code is available at https://github.com/huawei-noah/CV-Backbones/tree/master/wavemlp_pytorch and https://gitee.com/mindspore/models/tree/master/research/cv/wave_mlp.

In recent years, with the advances of generative models, many powerful face manipulation systems have been developed based on Deep Neural Networks (DNNs), called DeepFakes. If DeepFakes are not controlled timely and properly, they would become a real threat to both celebrities and ordinary people. Precautions such as adding perturbations to the source inputs will make DeepFake results look distorted from the perspective of human eyes. However, previous method doesn't explore whether the disrupted images can still spoof DeepFake detectors. This is critical for many applications where DeepFake detectors are used to discriminate between DeepFake data and real data due to the huge cost of examining a large amount of data manually. We argue that the detectors do not share a similar perspective as human eyes, which might still be spoofed by the disrupted data. Besides, the existing disruption methods rely on iteration-based perturbation generation algorithms, which is time-consuming. In this paper, we propose a novel DeepFake disruption algorithm called “DeepFake Disrupter”. By training a perturbation generator, we can add the human-imperceptible perturbations to source images that need to be protected without any backpropagation update. The DeepFake results of these protected source inputs would not only look unrealistic by the human eye but also can be distinguished by DeepFake detectors easily. For example, experimental results show that by adding our trained perturbations, fake images generated by StarGAN [5] can result in a 10 ∼ 20% increase in F1-score evaluated by various DeepFake detectors.

Learning with few labeled data has been a longstanding problem in the computer vision and machine learning research community. In this paper, we introduced a new semi-supervised learning framework, SimMatch, which simulta-neously considers semantic similarity and instance similarity. In SimMatch, the consistency regularization will be applied on both semantic-level and instance-level. The different augmented views of the same instance are encouraged to have the same class prediction and similar similarity re-lationship respected to other instances. Next, we instanti-ated a labeled memory buffer to fully leverage the ground truth labels on instance-level and bridge the gaps between the semantic and instance similarities. Finally, we proposed the unfolding and aggregation operation which allows these two similarities be isomorphically transformed with each other. In this way, the semantic and instance pseudo-labels can be mutually propagated to generate more high-quality and reliable matching targets. Extensive ex-perimental results demonstrate that SimMatch improves the performance of semi-supervised learning tasks across dif-ferent benchmark datasets and different settings. Notably, with 400 epochs of training, SimMatch achieves 67.2%, and 74.4% Top-1 Accuracy with 1% and 10% labeled examples on ImageNet, which significantly outperforms the baseline methods and is better than previous semi-supervised learning frameworks.

Filter pruning is essential for deploying a well-trained CNN model on edge computation devices with a target computation budget (e.g., FLOPs). Current filter pruning methods mainly focus on leveraging feature maps to analyze the importance of filters, and prune those with less impact on the value of the CNN’s loss function, thereby ignoring the variance of input batches to differences in sparse structure over the filters. In this paper, we propose a data-agnostic filter pruning method that uses an auxiliary network named Dagger module to induce pruning with the pre-trained weights as input. Besides, to help prune filters with a preset FLOPs constraint, we utilize an explicit FLOPs-aware regularisation mechanism to directly promote pruning filters toward the target FLOPs. Experimental results on CIFAR-10 and ImageNet datasets show that the proposed filter pruning method surpasses the state-of-the-art.

Pre-trained models for Natural Languages (NL) have been recently shown to transfer well to Programming Languages (PL) and largely benefit different intelligence code-related tasks, such as code search, clone detection, programming translation and code document generation. However, existing pre-trained methods for programming languages are mainly conducted by masked language modeling and next sentence prediction at token or graph levels. This restricted form limits their performance and transferability since PL and NL have different syntax rules and the downstream tasks require a multimodal representation. Here we introduce C3P, a Contrastive Code-Comment Pre-training approach, to solve various downstream tasks by pre-training the multi-representation features on both programming and natural syntax. The model encodes the code syntax and natural language description (comment) by two encoders and the encoded embeddings are projected into a multi-modal space for learning the latent representation. In the latent space, C3P jointly trains the code and comment encoders by the symmetric loss function, which aims to maximize the cosine similarity of the correct code-comment pairs while minimizing the similarity of unrelated pairs. We verify the empirical performance of the proposed pre-trained models on multiple downstream code-related tasks. The comprehensive experiments demonstrate that C3P outperforms previous work on the understanding tasks of code search and code clone, as well as the generation tasks of programming translation and document generation. Furthermore, we validate the transferability of C3P to the new programming language which is not seen in the pre-training stage. The results show our model surpasses all supervised methods and in some programming language cases even outperforms prior pre-trained approaches. Code is available at https://github.com/TerryPei/C3P.

Cognitive diagnosis models map observations onto psychological internal traits, which have been widely used to construct personalized educational assessments. However, the challenges of the latent ability representation, accurate performance with explanatory and cold-start problems caused by limited handcraft skills still exist. In this paper, we introduce a selfadaptive Attention Gate Cognitive Diagnosis Model (AGCDM) based on a multi-layer hierarchical structure. Specifically, in the first two layers, our model embeds the item response logs into a higher expressive latent space, and the specific self-attention mechanism captures the rich information from the inner-dependencies correlation among the examinee, quiz and knowledge. In the next layer, the gate mechanism addresses the challenge of noisy information (slip and guess) over psychological reasons. We verify the empirical performance of our model on different educational tasks and all comparisons study conducted on the same training recipe. On the item response prediction task, the experiment results demonstrate our hierarchical model achieves state-of-the-art performance. Through the estimation of adaptive learning, we also validate the model’s effectiveness in addressing the limited skill problem in real scenarios. The learning tracks show that our model empowered by the meta-learning algorithm can be adaptive faster than the base model to the new different skill distribution set. Furthermore, we present the ablation study on the hierarchical architecture and detailed hyperparameters selections. The comparison results demonstrate the attention blocks yield the scores improvement, and the estimation of the noisy errors shows the gate mechanism boosts the detection accuracy on the slip and guess. Code is available at https://github.com/TerryPei/AGCDM.

Recently, neural architectures with all Multi-layer Perceptrons (MLPs) have attracted great research interest from the computer vision community. However, the inefficient mixing of spatial-channel information causes MLP-like vision models to demand tremendous pre-training on large-scale datasets. This work solves the problem from a novel knowledge distillation perspective. We propose a novel Spatial-channel Token Distillation (STD) method, which improves the information mixing in the two dimensions by introducing distillation tokens to each of them. A mutual information regularization is further introduced to let distillation tokens focus on their specific dimensions and maximize the performance gain. Extensive experiments on ImageNet for several MLP-like architectures demonstrate that the proposed token distillation mechanism can efficiently improve the accuracy. For example, the proposed STD boosts the top-1 accuracy of Mixer-S16 on ImageNet from 73.8% to 75.7% without any costly pre-training on JFT-300M. When applied to stronger architectures, e.g. CycleMLP-B1 and CycleMLP-B2, STD can still harvest about 1.1% and 0.5% accuracy gains, respectively.

Currently, Vision Transformer (ViT) and its variants have demonstrated promising performance on various computer vision tasks. Nevertheless, task-irrelevant information such as background nuisance and noise in patch tokens would damage the performance of ViT-based models. In this paper, we develop Sufficient Vision Transformer (Suf-ViT) as a new solution to address this issue. In our research, we propose the Sufficiency-Blocks (S-Blocks) to be applied across the depth of Suf-ViT to disentangle and discard task-irrelevant information accurately. Besides, to boost the training of Suf-ViT, we formulate a Sufficient-Reduction Loss (SRLoss) leveraging the concept of Mutual Information (MI) that enables Suf-ViT to extract more reliable sufficient representations by removing task-irrelevant information. Extensive experiments on benchmark datasets such as ImageNet, ImageNet-C, and CIFAR-10 indicate that our method can achieve state-of-the-art or competing performance over other baseline methods. Codes are available at: https://github.com/zhicheng2T0/Sufficient-Vision-Transformer.git

Unlike existing knowledge distillation methods focus on the baseline settings, where the teacher models and training strategies are not that strong and competing as state-of-the-art approaches, this paper presents a method dubbed DIST to distill better from a stronger teacher. We empirically find that the discrepancy of predictions between the student and a stronger teacher may tend to be fairly severer. As a result, the exact match of predictions in KL divergence would disturb the training and make existing methods perform poorly. In this paper, we show that simply preserving the relations between the predictions of teacher and student would suffice, and propose a correlation-based loss to capture the intrinsic inter-class relations from the teacher explicitly. Besides, considering that different instances have different semantic similarities to each class, we also extend this relational match to the intra-class level. Our method is simple yet practical, and extensive experiments demonstrate that it adapts well to various architectures, model sizes and training strategies, and can achieve state-of-the-art performance consistently on image classification, object detection, and semantic segmentation tasks. Code is available at: https://github.com/hunto/DIST_KD.

Spatio-Temporal feature matching and alignment are essential for few-shot action recognition as they determine the coherence and effectiveness of the temporal patterns. Nevertheless, this process could be not reliable, especially when dealing with complex video scenarios. In this paper, we propose to improve the performance of matching and alignment from the end-to-end design of models. Our solution comes at two-folds. First, we encourage to enhance the extracted Spatio-Temporal representations from few-shot videos in the perspective of architectures. With this aim, we propose a specialized transformer search method for videos, thus the spatial and temporal attention can be well-organized and optimized for stronger feature representations. Second, we also design an efficient non-parametric spatio-temporal prototype alignment strategy to better handle the high variability of motion. In particular, a query-specific class prototype will be generated for each query sample and category, which can better match query sequences against all support sequences. By doing so, our method SST enjoys significant superiority over the benchmark UCF101 and HMDB51 datasets. For example, with no pretraining, our method achieves 17.1\% Top-1 accuracy improvement than the baseline TRX on UCF101 5-way 1-shot setting but with only 3x fewer FLOPs.

The vulnerability of deep neural networks has been widely found in various models as well as tasks where slight perturbations on the inputs could lead to incorrect predictions. These perturbed inputs are known as adversarial examples and one of the intriguing properties of them is Adversarial Transfersability, i.e. the capability of adversarial examples to fool other models. Traditionally, this transferability is always regarded as a critical threat to the defense against adversarial attacks, however, we argue that the network robustness can be significantly boosted by utilizing adversarial transferability from a new perspective. In this work, we first discuss the influence of different popular normalization layers on the adversarial transferability, and then provide both empirical evidence and theoretical analysis to shed light on the relationship between normalization types and transferability. Based on our theoretical analysis, we propose a simple yet effective module named Random Normalization Aggregation (RNA) which replaces the batch normalization layers in the networks and aggregates different selected normalization types to form a huge random space. Specifically, a random path is sampled during each inference procedure so that the network itself can be treated as an ensemble of a wide range of different models. Since the entire random space is designed with low adversarial transferability, it is difficult to perform effective attacks even when the network parameters are accessible. We conduct extensive experiments on various models and datasets, and demonstrate the strong superiority of proposed algorithm. The PyTorch code is available at https://github.com/UniSerj/Random-Norm-Aggregation and the MindSpore code is available at https://gitee.com/mindspore/models/tree/master/research/cv/RNA.

Light-weight convolutional neural networks (CNNs) are specially designed for applications on mobile devices with faster inference speed. The convolutional operation can only capture local information in a window region, which prevents performance from being further improved. Introducing self-attention into convolution can capture global information well, but it will largely encumber the actual speed. In this paper, we propose a hardware-friendly attention mechanism (dubbed DFC attention) and then present a new GhostNetV2 architecture for mobile applications. The proposed DFC attention is constructed based on fully-connected layers, which can not only execute fast on common hardware but also capture the dependence between long-range pixels. We further revisit the expressiveness bottleneck in previous GhostNet and propose to enhance expanded features produced by cheap operations with DFC attention, so that a GhostNetV2 block can aggregate local and long-range information simultaneously. Extensive experiments demonstrate the superiority of GhostNetV2 over existing architectures. For example, it achieves 75.3% top-1 accuracy on ImageNet with 167M FLOPs, significantly suppressing GhostNetV1 (74.5%) with a similar computational cost. The source code will be available at https://github.com/huawei-noah/Efficient-AI-Backbones/tree/master/ghostnetv2_pytorch and https://gitee.com/mindspore/models/tree/master/research/cv/ghostnetv2.

The generator in generative adversarial networks (GANs) is driven by a discriminator to produce high-quality images through an adversarial game. At the same time, the difficulty of reaching a stable generator has been increased. This paper focuses on non-adversarial generative networks that are trained in a plain manner without adversarial loss. The given limited number of real images could be insufficient to fully represent the real data distribution. We therefore investigate a set of distributions in a Wasserstein ball centred on the distribution induced by the training data and propose to optimize the generator over this Wasserstein ball. We theoretically discuss the solvability of the newly defined objective function and develop a tractable reformulation to learn the generator. The connections and differences between the proposed non-adversarial generative networks and GANs are analyzed. Experimental results on real-world datasets demonstrate that the proposed algorithm can effectively learn image generators in a non-adversarial approach, and the generated images are of comparable quality with those from GANs.

This paper introduces versatile filters to construct efficient convolutional neural networks that are widely used in various visual recognition tasks. Considering the demands of efficient deep learning techniques running on cost-effective hardware, a number of methods have been developed to learn compact neural networks. Most of these works aim to slim down filters in different ways, e.g., investigating small, sparse or quantized filters. In contrast, we treat filters from an additive perspective. A series of secondary filters can be derived from a primary filter with the help of binary masks. These secondary filters all inherit in the primary filter without occupying more storage, but once been unfolded in computation they could significantly enhance the capability of the filter by integrating information extracted from different receptive fields. Besides spatial versatile filters, we additionally investigate versatile filters from the channel perspective. Binary masks can be further customized for different primary filters under orthogonal constraints. We conduct theoretical analysis on network complexity and an efficient convolution scheme is introduced. Experimental results on benchmark datasets and neural networks demonstrate that our versatile filters are able to achieve comparable accuracy as that of original filters, but require less memory and computation cost.

In neural networks, developing regularization algorithms to settle overfitting is one of the major study areas. We propose a new approach for the regularization of neural networks by the local Rademacher complexity called LocalDrop. A new regularization function for both fully-connected networks (FCNs) and convolutional neural networks (CNNs), including drop rates and weight matrices, has been developed based on the proposed upper bound of the local Rademacher complexity by the strict mathematical deduction. The analyses of dropout in FCNs and DropBlock in CNNs with keep rate matrices in different layers are also included in the complexity analyses. With the new regularization function, we establish a two-stage procedure to obtain the optimal keep rate matrix and weight matrix to realize the whole training model. Extensive experiments have been conducted to demonstrate the effectiveness of LocalDrop in different models by comparing it with several algorithms and the effects of different hyperparameters on the final performances.

In this paper, we study the fine-grained categorization problem under the few-shot setting, i.e., each fine-grained class only contains a few labeled examples, termed Fine-Grained Few-Shot classification (FGFS). The core predicament in FGFS is the high intra-class variance yet low inter-class fluctuations in the dataset. In traditional fine-grained classification, the high intra-class variance can be somewhat relieved by conducting the supervised training on the abundant labeled samples. However, with few labeled examples, it is hard for the FGFS model to learn a robust class representation with the significantly higher intra-class variance. Moreover, the inter- and intra-class variance are closely related. The significant intra-class variance in FGFS often aggravates the low inter-class variance issue. To address the above challenges, we propose a Target-Oriented Alignment Network (TOAN) to tackle the FGFS problem from both intra- and inter-class perspective. To reduce the intra-class variance, we propose a target-oriented matching mechanism to reformulate the spatial features of each support image to match the query ones in the embedding space. To enhance the inter-class discrimination, we devise discriminative fine-grained features by integrating local compositional concept representations with the global second-order pooling. We conducted extensive experiments on four public datasets for fine-grained categorization, and the results show the proposed TOAN obtains the state-of-the-art.

Many regression tasks encounter an asymmetric distribution of information between training and testing phases where the additional information available in training, the so-called privileged information (PI), is often inaccessible in testing. In practice, the privileged information in training data might be expressed in different formats, such as continuous, ordinal, or binary values. However, most the existing learning using privileged information (LUPI) paradigms primarily deal with the continuous form of PI, preventing them from managing variational PI, which motivates this research. Therefore, in this paper, we propose a unified framework to systematically address the aforementioned three forms of privileged information. The proposed V-SVR+ method integrates continuous, ordinal, and binary PI into the learning process of support vector regression (SVR) via three losses. For continuous privileged information, we define a linear correcting (slack) function in the privileged information space to estimate slack variables in the standard SVR method using privileged information. For the ordinal relations of privileged information, we first rank the privileged information and then, regard this ordinal privileged information as auxiliary information used in the learning process of the SVR model. For the binary or Boolean privileged information, we infer a probabilistic dependency between the privileged information and labels from the summarized privileged information knowledge. Then, we transfer the privileged information knowledge to constraints and form a constrained optimization problem. We evaluate the proposed method in three applications: music emotion recognition from songs with the help of implicit information about music elements judged by composers; multiple object recognition from images with the help of implicit information about the object’s importance conveyed by the list of manually annotated image tags; and photo aesthetic assessment enhanced by high-level aesthetic attributes hidden in photos. Experiment results demonstrate that the proposed methods are superior to the classic learning paradigm when solving practical problems.

The predicament in semi-supervised few-shot learning (SSFSL) is to maximize the value of the extra unlabeled data to boost the few-shot learner. In this paper, we propose a Poisson Transfer Network (PTN) to mine the unlabeled information for SSFSL from two aspects. First, the Poisson Merriman–Bence–Osher (MBO) model builds a bridge for the communications between labeled and unlabeled examples. This model serves as a more stable and informative classifier than traditional graph-based SSFSL methods in the message-passing process of the labels. Second, the extra unlabeled samples are employed to transfer the knowledge from base classes to novel classes through contrastive learning. Specifically, we force the augmented positive pairs close while push the negative ones distant. Our contrastive transfer scheme implicitly learns the novel-class embeddings to alleviate the over-fitting problem on the few labeled data. Thus, we can mitigate the degeneration of embedding generality in novel classes. Extensive experiments indicate that PTN outperforms the state-of-the-art few-shot and SSFSL models on miniImageNet and tieredImageNet benchmark datasets.

Relying on the diverse graph convolution operations that have emerged in recent years, graph neural networks (GNNs) are shown to be powerful to deal with high-dimensional non-Euclidean domains, such as social networks or citation networks. Despite the tremendous human efforts been taken to explore new graph convolution operations, there are a few attempts to automatically search operations in GNNs. The search space of GNNs is significantly larger than that of CNNs, because of diverse components in the message-passing of GNNs. This, therefore, prevents the straightforward application of classical NAS methods for GNNs. In this work, we propose a novel dynamic one-shot search space for multi-branch neural architectures of GNNs. The dynamic search space maintains a subset of the large search space along with a set of importance weights for operation candidates in the subset as the architecture parameters. After each iteration, the subset is pruned by removing candidates with low importance weights and is expanded with new operations. The dynamic subsets of operation candidates are not uniform but is individual for each edge in the computation graph of the neural architecture, which can ensure the diversity of operations in the final architecture is as competitive as direct search in the large search space. Our experiments of semi-supervised and supervised node classification on citation networks, including Cora, Citeseer, and Pubmed, demonstrate that our method outperforms the current state-of-the-art manually designed architectures and reaches competitive performance to existing GNN NAS approaches with up to 10 times of speedup.

Despite the remarkable empirical performance of deep learning models, their vulnerability to adversarial examples has been revealed in many studies. They are prone to make a susceptible prediction to the input with imperceptible adversarial perturbation. Although recent works have remarkably improved the model's robustness under the adversarial training strategy, an evident gap between the natural accuracy and adversarial robustness inevitably exists. In order to mitigate this problem, in this paper, we assume that the robust and non-robust representations are two basic ingredients entangled in the integral representation. For achieving adversarial robustness, the robust representations of natural and adversarial examples should be disentangled from the non-robust part and the alignment of the robust representations can bridge the gap between accuracy and robustness. Inspired by this motivation, we propose a novel defense method called Deep Robust Representation Disentanglement Network (DRRDN). Specifically, DRRDN employs a disentangler to extract and align the robust representations from both adversarial and natural examples. Theoretical analysis guarantees the mitigation of the trade-off between robustness and accuracy with good disentanglement and alignment performance. Experimental results on benchmark datasets finally demonstrate the empirical superiority of our method.

As the computing power of modern hardware is increasing strongly, pre-trained deep learning models (e.g., BERT, GPT-3) learned on large-scale datasets have shown their effectiveness over conventional methods. The big progress is mainly contributed to the representation ability of transformer and its variant architectures. In this paper, we study the low-level computer vision task (e.g., denoising, super-resolution and deraining) and develop a new pre-trained model, namely, image processing transformer (IPT). To maximally excavate the capability of transformer, we present to utilize the well-known ImageNet benchmark for generating a large amount of corrupted image pairs. The IPT model is trained on these images with multi-heads and multi-tails. In addition, the contrastive learning is introduced for well adapting to different image processing tasks. The pre-trained model can therefore efficiently employed on desired task after fine-tuning. With only one pre-trained model, IPT outperforms the current state-of-the-art methods on various low-level benchmarks. Code is available at https://github.com/huawei-noah/Pretrained-IPT and https://gitee.com/mindspore/mindspore/tree/master/model_zoo/research/cv/IPT

Data-free learning for student networks is a new paradigm for solving users’ anxiety caused by the privacy problem of using original training data. Since the architectures of modern convolutional neural networks (CNNs) are compact and sophisticated, the alternative images or meta-data generated from the teacher network are often broken. Thus, the student network cannot achieve the comparable performance to that of the pre-trained teacher network especially on the large-scale image dataset. Different to previous works, we present to maximally utilize the massive available unlabeled data in the wild. Specifically, we first thoroughly analyze the output differences between teacher and student network on the original data and develop a data collection method. Then, a noisy knowledge distillation algorithm is proposed for achieving the performance of the student network. In practice, an adaptation matrix is learned with the student network for correcting the label noise produced by the teacher network on the collected unlabeled images. The effectiveness of our DFND (Data-Free Noisy Distillation) method is then verified on several benchmarks to demonstrate its superiority over state-of-the-art data-free distillation methods. Experiments on various datasets demonstrate that the student networks learned by the proposed method can achieve comparable performance with those using the original dataset. Code is available at https://github.com/huawei-noah/Data-Efficient-Model-Compression

Knowledge distillation is a widely used paradigm for inheriting information from a complicated teacher network to a compact student network and maintaining the strong performance. Different from image classification, object detectors are much more sophisticated with multiple loss functions in which features that semantic information rely on are tangled. In this paper, we point out that the information of features derived from regions excluding objects are also essential for distilling the student detector, which is usually ignored in existing approaches. In addition, we elucidate that features from different regions should be assigned with different importance during distillation. To this end, we present a novel distillation algorithm via decoupled features (DeFeat) for learning a better student detector. Specifically, two levels of decoupled features will be processed for embedding useful information into the student, i.e., decoupled features from neck and decoupled proposals from classification head. Extensive experiments on various detectors with different backbones show that the proposed DeFeat is able to surpass the state-of-the-art distillation methods for object detection. For example, DeFeat improves ResNet50 based Faster R-CNN from 37.4% to 40.9% mAP, and improves ResNet50 based RetinaNet from 36.5% to 39.7% mAP on COCO benchmark. Code will be released1,2.

Deep learning based object detection approaches have achieved great progress with the benefit from large amount of labeled images. However, image annotation remains a laborious, time-consuming and error-prone process. To further improve the performance of detectors, we seek to exploit all available labeled data and excavate useful samples from massive unlabeled images in the wild, which is rarely discussed before. In this paper, we present a positive-unlabeled learning based scheme to expand training data by purifying valuable images from massive unlabeled ones, where the original training data are viewed as positive data and the unlabeled images in the wild are unlabeled data. To effectively utilized these purified data, we propose a self-distillation algorithm based on hint learning and ground truth bounded knowledge distillation. Experimental results verify that the proposed positive-unlabeled data purification can strengthen the original detector by mining the massive unlabeled data. In particular, our method boosts the mAP of FPN by +2.0% on COCO benchmark.

This paper studies the single image super-resolution problem using adder neural networks (AdderNets). Com-pared with convolutional neural networks, AdderNets utilize additions to calculate the output features thus avoid massive energy consumptions of conventional multiplications. However, it is very hard to directly inherit the existing success of AdderNets on large-scale image classification to the image super-resolution task due to the different calculation paradigm. Specifically, the adder operation cannot easily learn the identity mapping, which is essential for image processing tasks. In addition, the functionality of high-pass filters cannot be ensured by AdderNets. To this end, we thoroughly analyze the relationship between an adder operation and the identity mapping and insert shortcuts to enhance the performance of SR models using adder networks. Then, we develop a learnable power activation for adjusting the feature distribution and refining details. Experiments conducted on several benchmark models and datasets demonstrate that, our image super-resolution models using AdderNets can achieve comparable performance and visual quality to that of their CNN baselines with an about 2.5× reduction on the energy consumption. The codes are available at: https://github.com/huawei-noah/AdderNet.

One-shot neural architecture search (NAS) methods significantly reduce the search cost by considering the whole search space as one network, which only needs to be trained once. However, current methods select each operation independently without considering previous layers. Besides, the historical information obtained with huge computation costs is usually used only once and then discarded. In this paper, we introduce a sampling strategy based on Monte Carlo tree search (MCTS) with the search space modeled as a Monte Carlo tree (MCT), which captures the dependency among layers. Furthermore, intermediate results are stored in the MCT for future decisions and a better exploration-exploitation balance. Concretely, MCT is updated using the training loss as a reward to the architecture performance; for accurately evaluating the numerous nodes, we propose node communication and hierarchical node selection methods in the training and search stages, respectively, making better uses of the operation rewards and hierarchical information. Moreover, for a fair comparison of different NAS methods, we construct an open-source NAS benchmark of a macro search space evaluated on CIFAR-10, namely NAS-Bench-Macro. Extensive experiments on NAS-Bench-Macro and ImageNet demonstrate that our method significantly improves search efficiency and performance. For example, by only searching 20 architectures, our obtained architecture achieves 78.0% top-1 accuracy with 442M FLOPs on ImageNet. Code (Benchmark) is available at: https://github.com/xiusu/NAS-Bench-Macro.

Searching for a more compact network width recently serves as an effective way of channel pruning for the deployment of convolutional neural networks (CNNs) under hardware constraints. To fulfill the searching, a one-shot supernet is usually leveraged to efficiently evaluate the performance w.r.t. different network widths. However, current methods mainly follow a unilaterally augmented (UA) principle for the evaluation of each width, which induces the training unfairness of channels in supernet. In this paper, we introduce a new supernet called Bilaterally Coupled Network (BCNet) to address this issue. In BCNet, each channel is fairly trained and responsible for the same amount of network widths, thus each network width can be evaluated more accurately. Besides, we leverage a stochastic complementary strategy for training the BCNet, and propose a prior initial population sampling method to boost the performance of the evolutionary search. Extensive experiments on benchmark CIFAR-10 and ImageNet datasets indicate that our method can achieve state-of-the-art or competing performance over other baseline methods. Moreover, our method turns out to further boost the performance of NAS models by refining their network widths. For example, with the same FLOPs budget, our obtained EfficientNet-B0 achieves 77.36% Top-1 accuracy on ImageNet dataset, surpassing the performance of original setting by 0.48%.

Neural network pruning is an essential approach for reducing the computational complexity of deep models so that they can be well deployed on resource-limited devices. Compared with conventional methods, the recently developed dynamic pruning methods determine redundant filters variant to each input instance which achieves higher acceleration. Most of the existing methods discover effective subnetworks for each instance independently and do not utilize the relationship between different inputs. To maximally excavate redundancy in the given network architecture, this paper proposes a new paradigm that dynamically removes redundant filters by embedding the manifold information of all instances into the space of pruned networks (dubbed as ManiDP). We first investigate the recognition complexity and feature similarity between images in the training set. Then, the manifold relationship between instances and the pruned sub-networks will be aligned in the training procedure. The effectiveness of the proposed method is verified on several benchmarks, which shows better performance in terms of both accuracy and computational cost compared to the state-of-the-art methods. For example, our method can reduce 55.3% FLOPs of ResNet-34 with only 0.57% top-1 accuracy degradation on ImageNet. The code will be available at https://github.com/huawei-noah/Pruning/tree/master/ManiDP.

An effective and efficient architecture performance evaluation scheme is essential for the success of Neural Architecture Search (NAS). To save computational cost, most of existing NAS algorithms often train and evaluate intermediate neural architectures on a small proxy dataset with limited training epochs. But it is difficult to expect an accurate performance estimation of an architecture in such a coarse evaluation way. This paper advocates a new neural architecture evaluation scheme, which aims to determine which architecture would perform better instead of accurately predict the absolute architecture performance. Therefore, we propose a relativistic architecture performance predictor in NAS (ReNAS). We encode neural architectures into feature tensors, and further refining the representations with the predictor. The proposed relativistic performance predictor can be deployed in discrete searching methods to search for the desired architectures without additional evaluation. Experimental results on NAS-Bench-101 dataset suggests that, sampling 424 (0.1% of the entire search space) neural architectures and their corresponding validation performance is already enough for learning an accurate architecture performance predictor. The accuracies of our searched neural architectures on NAS-Bench-101 and NAS-Bench-201 datasets are higher than that of the state-of-the-art methods and show the priority of the proposed method.

Neural Architecture Search (NAS) aims to automatically discover optimal architectures. In this paper, we propose an hourglass-inspired approach (HourNAS) for extremely fast NAS. It is motivated by the fact that the effects of the architecture often proceed from the vital few blocks. Acting like the narrow neck of an hourglass, vital blocks in the guaranteed path from the input to the output of a deep neural network restrict the information flow and influence the network accuracy. The other blocks occupy the major volume of the network and determine the overall network complexity, corresponding to the bulbs of an hourglass. To achieve an extremely fast NAS while preserving the high accuracy, we propose to identify the vital blocks and make them the priority in the architecture search. The search space of those non-vital blocks is further shrunk to only cover the candidates that are affordable under the computational resource constraints. Experimental results on ImageNet show that only using 3 hours (0.1 days) with one GPU, our HourNAS can search an architecture that achieves a 77.0% Top-1 accuracy, which outperforms the state-of-the-art methods.

Capturing interpretable variations has long been one of the goals in disentanglement learning. However, unlike the independence assumption, interpretability has rarely been exploited to encourage disentanglement in the unsupervised setting. In this paper, we examine the interpretability of disentangled representations by investigating two questions: where to be interpreted and what to be interpreted? A latent code is easily to be interpreted if it would consistently impact a certain subarea of the resulting generated image. We thus propose to learn a spatial mask to localize the effect of each individual latent dimension. On the other hand, interpretability usually comes from latent dimensions that capture simple and basic variations in data. We thus impose a perturbation on a certain dimension of the latent code, and expect to identify the perturbation along this dimension from the generated images so that the encoding of simple variations can be enforced. Additionally, we develop an unsupervised model selection method, which accumulates perceptual distance scores along axes in the latent space. On various datasets, our models can learn high-quality disentangled representations without supervision, showing the proposed modeling of interpretability is an effective proxy for achieving unsupervised disentanglement.

Deep learning based methods, especially convolutional neural networks (CNNs) have been successfully applied in the field of single image super-resolution (SISR). To obtain better fidelity and visual quality, most of existing networks are of heavy design with massive computation. However, the computation resources of modern mobile devices are limited, which cannot easily support the expensive cost. To this end, this paper explores a novel frequency-aware dynamic network for dividing the input into multiple parts according to its coefficients in the discrete cosine transform (DCT) domain. In practice, the high-frequency part will be processed using expensive operations and the lower-frequency part is assigned with cheap operations to relieve the computation burden. Since pixels or image patches belong to low-frequency areas contain relatively few textural details, this dynamic network will not affect the quality of resulting super-resolution images. In addition, we embed predictors into the proposed dynamic network to end-to-end fine-tune the handcrafted frequency-aware masks. Extensive experiments conducted on benchmark SISR models and datasets show that the frequency-aware dynamic network can be employed for various SISR neural architectures to obtain the better tradeoff between visual quality and computational complexity. For instance, we can reduce the FLOPs of SR models by approximate 50% while preserving state-of-the-art SISR performance.

Unsupervised visual representation learning has gained much attention from the computer vision community because of the recent achievement of contrastive learning. Most of the existing contrastive learning frameworks adopt the instance discrimination as the pretext task, which treating every single instance as a different class. However, such method will inevitably cause class collision problems, which hurts the quality of the learned representation. Motivated by this observation, we introduced a weakly supervised contrastive learning framework (WCL) to tackle this issue. Specifically, our proposed framework is based on two projection heads, one of which will perform the regular instance discrimination task. The other head will use a graph-based method to explore similar samples and generate a weak label, then perform a supervised contrastive learning task based on the weak label to pull the similar images closer. We further introduced a K-Nearest Neighbor based multi-crop strategy to expand the number of positive samples. Extensive experimental results demonstrate WCL improves the quality of self-supervised representations across different datasets. Notably, we get a new state-of-the-art result for semi-supervised learning. With only 1% and 10% labeled examples, WCL achieves 65% and 72% ImageNet Top-1 Accuracy using ResNet50, which is even higher than SimCLRv2 with ResNet101.

Recent researches indicate that the impact of adversarial perturbations on deep learning models is reflected not only on the alteration of predicted labels but also on the distortion of data structure in the representation space. Significant improvement of the model’s adversarial robustness can be achieved by reforming the structure-aware representation distortion. Current methods generally utilize the one-to-one representation alignment or the triplet information between the positive and negative pairs. However, in this paper, we show that the representation structure of the natural and adversarial examples cannot be well and stably captured if we only focus on a localized range of contrastive examples. To achieve better and more stable adversarial robustness, we propose to adjust the adversarial distortion of representation structure by using Massive Contrastive Adversaries (MCA). Inspired by the Noise-Contrastive Estimation (NCE), MCA exploits the contrastive information by employing m negative instances. Compared with existing methods, our method recruits a much wider range of negative examples per update, so a better and more stable representation relationship between the natural and adversarial examples can be captured. Theoretical analysis shows that the proposed MCA inherently maximizes a lower bound of the mutual information (MI) between the representations of the natural and adversarial examples. Empirical experiments on benchmark datasets demonstrate that MCA can achieve better and more stable intra-class compactness and inter-class divergence, which further induces better adversarial robustness.

In one-shot weight sharing for NAS, the weights of each operation (at each layer) are supposed to be identical for all architectures (paths) in the supernet. However, this rules out the possibility of adjusting operation weights to cater for different paths, which limits the reliability of the evaluation results. In this paper, instead of counting on a single supernet, we introduce $K$-shot supernets and take their weights for each operation as a dictionary. The operation weight for each path is represented as a convex combination of items in a dictionary with a simplex code. This enables a matrix approximation of the stand-alone weight matrix with a higher rank ($K>1$). A \textit{simplex-net} is introduced to produce architecture-customized code for each path. As a result, all paths can adaptively learn how to share weights in the $K$-shot supernets and acquire corresponding weights for better evaluation. $K$-shot supernets and simplex-net can be iteratively trained, and we further extend the search to the channel dimension. Extensive experiments on benchmark datasets validate that K-shot NAS significantly improves the evaluation accuracy of paths and thus brings in impressive performance improvements.

This paper investigates how to weight imperfect expert demonstrations for generative adversarial imitation learning (GAIL). The agent is expected to perform behaviors demonstrated by experts. But in many applications, experts could also make mistakes and their demonstrations would mislead or slow the learning process of the agent. Recently, existing methods for imitation learning from imperfect demonstrations mostly focus on using the preference or confidence scores to distinguish imperfect demonstrations. However, these auxiliary information needs to be collected with the help of an oracle, which is usually hard and expensive to afford in practice. In contrast, this paper proposes a method of learning to weight imperfect demonstrations in GAIL without imposing extensive prior information. We provide a rigorous mathematical analysis, presenting that the weights of demonstrations can be exactly determined by combining the discriminator and agent policy in GAIL. Theoretical analysis suggests that with the estimated weights the agent can learn a better policy beyond those plain expert demonstrations. Experiments in the Mujoco and Atari environments demonstrate that the proposed algorithm outperforms baseline methods in handling imperfect expert demonstrations.

We view disentanglement learning as discovering an underlying structure that equivariantly reflects the factorized variations shown in data. Traditionally, such a structure is fixed to be a vector space with data variations represented by translations along individual latent dimensions. We argue this simple structure is suboptimal since it requires the model to learn to discard the properties (e.g. different scales of changes, different levels of abstractness) of data variations, which is an extra work than equivariance learning. Instead, we propose to encode the data variations with groups, a structure not only can equivariantly represent variations, but can also be adaptively optimized to preserve the properties of data variations. Considering it is hard to conduct training on group structures, we focus on Lie groups and adopt a parameterization using Lie algebra. Based on the parameterization, some disentanglement learning constraints are naturally derived. A simple model named Commutative Lie Group VAE is introduced to realize the group-based disentanglement learning. Experiments show that our model can effectively learn disentangled representations without supervision, and can achieve state-of-the-art performance without extra constraints.

The agent in imitation learning (IL) is expected to mimic the behavior of the expert. Its performance relies highly on the quality of given expert demonstrations. However, the assumption that collected demonstrations are optimal cannot always hold in real-world tasks, which would seriously influence the performance of the learned agent. In this paper, we propose a robust method within the framework of Generative Adversarial Imitation Learning (GAIL) to address imperfect demonstration issue, in which good demonstrations can be adaptively selected for training while bad demonstrations are abandoned. Specifically, a binary weight is assigned to each expert demonstration to indicate whether to select it for training. The reward function in GAIL is employed to determine this weight (i.e. higher reward results in higher weight). Compared to some existing solutions that require some auxiliary information about this weight, we set up the connection between weight and model so that we can jointly optimize GAIL and learn the latent weight. Besides hard binary weighting, we also propose a soft weighting scheme. Experiments in the Mujoco demonstrate the proposed method outperforms other GAIL-based methods when dealing with imperfect demonstrations.

The first international Workshop on Programming Language Processing presents interdisciplinary contributions that address programming language procession problems with machine learning and data mining techniques. Recently, there are lots of successful natural language processing methods. But the mining of programming languages could not exactly follow the manner of natural language processing. The difference between natural language and programming language brings in new research challenges and opportunities. The workshop will bring together researchers from machine learning, data mining and software engineering to discuss and debate the path forward for mining the value of programming languages.

How to mine the knowledge in the pretrained models is of significance in achieving more promising performance, since practitioners have access to many pretrained models easily. This Workshop on Model Mining aims to investigate more diverse and advanced manners in mining knowledge within models, which tends to leverage the pretrained models more wisely, elegantly and systematically. There are many topics related to this workshop, such as distilling a lightweight model from a well-trained heavy model via teacher-student paradigm, and boosting the performance of the model by carefully designing the predecessor tasks, e.g., pre-training, self-supervised and contrastive learning. Model mining as a special way of data mining is relevant to SIGKDD, and its audience including researchers and engineers will benefit a lot for designing more advanced algorithms for their tasks.

Adder neural networks (AdderNets) have shown impressive performance on image classification with only addition operations, which are more energy efficient than traditional convolutional neural networks built with multiplications. Compared with classification, there is a strong demand on reducing the energy consumption of modern object detectors via AdderNets for real-world applications such as autonomous driving and face detection. In this paper, we present an empirical study of AdderNets for object detection. We first reveal that the batch normalization statistics in the pre-trained adder backbone should not be frozen, since the relatively large feature variance of AdderNets. Moreover, we insert more shortcut connections in the neck part and design a new feature fusion architecture for avoiding the sparse features of adder layers. We present extensive ablation studies to explore several design choices of adder detectors. Comparisons with state-of-the-arts are conducted on COCO and PASCAL VOC benchmarks. Specifically, the proposed Adder FCOS achieves a 37.8% AP on the COCO val set, demonstrating comparable performance to that of the convolutional counterpart with an about $1.4\times$ energy reduction.

Adder neural network (AdderNet) replaces the original convolutions with massive multiplications by cheap additions while achieving comparable performance thus yields a series of energy-efficient neural networks. Compared with convolutional neural networks (CNNs), the training of AdderNets is much more sophisticated including several techniques for adjusting gradient and batch normalization. In addition, variances of both weights and activations in resulting adder networks are very enormous which limits its performance and the potential for applying to other tasks. To enhance the stability and robustness of AdderNets, we first thoroughly analyze the variance estimation of weight parameters and output features of an arbitrary adder layer. Then, we develop a weight normalization scheme for adaptively optimizing the weight distribution of AdderNets during the training procedure, which can reduce the perturbation on running mean and variance in batch normalization layers. Meanwhile, the proposed weight normalization can also be utilized to enhance the adversarial robustness of resulting networks. Experiments conducted on several benchmarks demonstrate the superiority of the proposed approach for generating AdderNets with higher performance.

Adder neural networks (ANNs) are designed for low energy cost which replace expensive multiplications in convolutional neural networks (CNNs) with cheaper additions to yield energy-efficient neural networks and hardware accelerations. Although ANNs achieve satisfactory efficiency, there exist gaps between ANNs and CNNs where the accuracy of ANNs can hardly be compared to CNNs without the assistance of other training tricks, such as knowledge distillation. The inherent discrepancy lies in the similarity measurement between filters and features, however how to alleviate this difference remains unexplored. To locate the potential problem of ANNs, we focus on the property difference due to similarity measurement. We demonstrate that unordered heavy tails in ANNs could be the key component which prevents ANNs from achieving superior classification performance since fatter tails tend to overlap in feature space. Through pre-defining Multivariate Skew Laplace distributions and embedding feature distributions into the loss function, ANN features can be fully controlled and designed for various properties. We further present a novel method for tackling existing heavy tails in ANNs with only a modification of classifier where ANN features are clustered with their tails well-formulated through proposed angle-based constraint on the distribution parameters to encourage high diversity of tails. Experiments conducted on several benchmarks and comparison with other distributions demonstrate the effectiveness of proposed approach for boosting the performance of ANNs.

With the tremendous advances in the architecture and scale of convolutional neural networks (CNNs) over the past few decades, they can easily reach or even exceed the performance of humans in certain tasks. However, a recently discovered shortcoming of CNNs is that they are vulnerable to adversarial attacks. Although the adversarial robustness of CNNs can be improved by adversarial training, there is a trade-off between standard accuracy and adversarial robustness. From the neural architecture perspective, this paper aims to improve the adversarial robustness of the backbone CNNs that have a satisfactory accuracy. Under a minimal computational overhead, the introduction of a dilation architecture is expected to be friendly with the standard performance of the backbone CNN while pursuing adversarial robustness. Theoretical analyses on the standard and adversarial error bounds naturally motivate the proposed neural architecture dilation algorithm. Experimental results on real-world datasets and benchmark neural networks demonstrate the effectiveness of the proposed algorithm to balance the accuracy and adversarial robustness.

Transformer models have achieved great progress on computer vision tasks recently. The rapid development of vision transformers is mainly contributed by their high representation ability for extracting informative features from input images. However, the mainstream transformer models are designed with deep architectures, and the feature diversity will be continuously reduced as the depth increases, \ie, feature collapse. In this paper, we theoretically analyze the feature collapse phenomenon and study the relationship between shortcuts and feature diversity in these transformer models. Then, we present an augmented shortcut scheme, which inserts additional paths with learnable parameters in parallel on the original shortcuts. To save the computational costs, we further explore an efficient approach that uses the block-circulant projection to implement augmented shortcuts. Extensive experiments conducted on benchmark datasets demonstrate the effectiveness of the proposed method, which brings about 1% accuracy increase of the state-of-the-art visual transformers without obviously increasing their parameters and FLOPs.

Binary neural networks (BNNs) represent original full-precision weights and activations into 1-bit with sign function. Since the gradient of the conventional sign function is almost zero everywhere which cannot be used for back-propagation, several attempts have been proposed to alleviate the optimization difficulty by using approximate gradient. However, those approximations corrupt the main direction of factual gradient. To this end, we propose to estimate the gradient of sign function in the Fourier frequency domain using the combination of sine functions for training BNNs, namely frequency domain approximation (FDA). The proposed approach does not affect the low-frequency information of the original sign function which occupies most of the overall energy, and high-frequency coefficients will be ignored to avoid the huge computational overhead. In addition, we embed a noise adaptation module into the training phase to compensate the approximation error. The experiments on several benchmark datasets and neural architectures illustrate that the binary network learned using our method achieves the state-of-the-art accuracy. Code will be available at https://gitee.com/mindspore/models/tree/master/research/cv/FDA-BNN.

Self-supervised Learning (SSL) including the mainstream contrastive learning has achieved great success in learning visual representations without data annotations. However, most of methods mainly focus on the instance level information (\ie, the different augmented images of the same instance should have the same feature or cluster into the same class), but there is a lack of attention on the relationships between different instances. In this paper, we introduced a novel SSL paradigm, which we term as relational self-supervised learning (ReSSL) framework that learns representations by modeling the relationship between different instances. Specifically, our proposed method employs sharpened distribution of pairwise similarities among different instances as \textit{relation} metric, which is thus utilized to match the feature embeddings of different augmentations. Moreover, to boost the performance, we argue that weak augmentations matter to represent a more reliable relation, and leverage momentum strategy for practical efficiency. Experimental results show that our proposed ReSSL significantly outperforms the previous state-of-the-art algorithms in terms of both performance and training efficiency.

Unsupervised open-set domain adaptation (UODA) is a realistic problem where unlabeled target data contain unknown classes. Prior methods rely on the coexistence of both source and target domain data to perform domain alignment, which greatly limits their applications when source domain data are restricted due to privacy concerns. In this paper we address the challenging hypothesis transfer setting for UODA, where data from source domain are no longer available during adaptation on target domain. Specifically, we propose to use pseudo-labels and a novel consistency regularization on target data, where using conventional formulations fails in this open-set setting. Firstly, our method discovers confident predictions on target domain and performs classification with pseudo-labels. Then we enforce the model to output consistent and definite predictions on semantically similar transformed inputs, discovering all latent class semantics. As a result, unlabeled data can be classified into discriminative classes coincided with either source classes or unknown classes. We theoretically prove that under perfect semantic transformation, the proposed objective that enforces consistency can recover the information of true labels in prediction. Experimental results show that our model outperforms state-of-the-art methods on UODA benchmarks.

Deep convolutional neural networks have been widely used in numerous applications, but their demanding storage and computational resource requirements prevent their applications on mobile devices. Knowledge distillation aims to optimize a portable student network by taking the knowledge from a well-trained heavy teacher network. Traditional teacher-student-based methods used to rely on additional fully connected layers to bridge intermediate layers of teacher and student networks, which brings in a large number of auxiliary parameters. In contrast, this article aims to propagate information from teacher to student without introducing new variables that need to be optimized. We regard the teacher-student paradigm from a new perspective of feature embedding. By introducing the locality preserving loss, the student network is encouraged to generate the low-dimensional features that could inherit intrinsic properties of their corresponding high-dimensional features from the teacher network. The resulting portable network, thus, can naturally maintain the performance as that of the teacher network. Theoretical analysis is provided to justify the lower computation complexity of the proposed method. Experiments on benchmark data sets and well-trained networks suggest that the proposed algorithm is superior to state-of-the-art teacher-student learning methods in terms of computational and storage complexity.

Despite Generative Adversarial Networks (GANs) have been widely used in various image-to-image translation tasks, they can be hardly applied on mobile devices due to their heavy computation and storage cost. Traditional network compression methods focus on visually recognition tasks, but never deal with generation tasks. Inspired by knowledge distillation, a student generator of fewer parameters is trained by inheriting the low-level and high-level information from the original heavy teacher generator. To promote the capability of student generator, we include a student discriminator to measure the distances between real images, and images generated by student and teacher generators. An adversarial learning process is therefore established to optimize student generator and student discriminator. Qualitative and quantitative analysis by conducting experiments on benchmark datasets demonstrate that the proposed method can learn portable generative models with strong performance.

Recently, the teacher-student learning paradigm has drawn much attention in compressing neural networks on low-end edge devices, such as mobile phones and wearable watches. Current algorithms mainly assume the complete dataset for the teacher network is also available for the training of the student network. However, for real-world scenarios, users may only have access to part of training examples due to commercial profits or data privacy, and severe over-fitting issues would happen as a result. In this paper, we tackle the challenge of learning student networks with few data by investigating the ground-truth data-generating distribution underlying these few data. Taking Wasserstein distance as the measurement, we assume this ideal data distribution lies in a neighborhood of the discrete empirical distribution induced by the training examples. Thus we propose to safely optimize the worst-case cost within this neighborhood to boost the generalization. Furthermore, with theoretical analysis, we derive a novel and easy-to-implement loss for training the student network in an end-to-end fashion. Experimental results on benchmark datasets validate the effectiveness of our proposed method.

Although remarkable progress has been made on single image super-resolution due to the revival of deep convolutional neural networks, deep learning methods are confronted with the challenges of computation and memory consumption in practice, especially for mobile devices. Focusing on this issue, we propose an efficient residual dense block search algorithm with multiple objectives to hunt for fast, lightweight and accurate networks for image super-resolution. Firstly, to accelerate super-resolution network, we exploit the variation of feature scale adequately with the proposed efficient residual dense blocks. In the proposed evolutionary algorithm, the locations of pooling and upsampling operator are searched automatically. Secondly, network architecture is evolved with the guidance of block credits to acquire accurate super-resolution network. The block credit reflects the effect of current block and is earned during model evaluation process. It guides the evolution by weighing the sampling probability of mutation to favor admirable blocks. Extensive experimental results demonstrate the effectiveness of the proposed searching method and the found efficient super-resolution models achieve better performance than the state-of-the-art methods with limited number of parameters and FLOPs.

Deep neural networks often consist of a great number of trainable parameters for extracting powerful features from given datasets. One one hand, massive trainable parameters significantly enhance the performance of these deep networks. One the other hand, they bring the problem of over-fitting. To this end, dropout based methods disable some elements in the output feature maps during the training phase for reducing the co-adaptation of neurons. Although the generalization ability of the resulting models can be enhanced by these approaches, the conventional binary dropout is not the optimal solution. Therefore, we investigate the empirical Rademacher complexity related to intermediate layers of deep neural networks and propose a feature distortion method for addressing the aforementioned problem. In the training period, randomly selected elements in the feature maps will be replaced with specific values by exploiting the generalization error bound. The superiority of the proposed feature map distortion for producing deep neural network with higher testing performance is analyzed and demonstrated on several benchmark image datasets.

Channel pruning is effective in compressing the pretrained CNNs for their deployment on low-end edge devices. Most existing methods independently prune some of the original channels and need the complete original dataset to fix the performance drop after pruning. However, due to commercial protection or data privacy, users may only have access to a tiny portion of training examples, which could be insufficient for the performance recovery. In this paper, for pruning with limited data, we propose to use all original filters to directly develop new compact filters, named reborn filters, so that all useful structure priors in the original filters can be well preserved into the pruned networks, alleviating the performance drop accordingly. During training, reborn filters can be easily implemented via 1×1 convolutional layers and then be fused in the inference stage for acceleration. Based on reborn filters, the proposed channel pruning algorithm shows its effectiveness and superiority on extensive experiments.

Graph convolutional models have gained impressive successes on skeleton based human action recognition task. As graph convolution is a local operation, it cannot fully investigate non-local joints that could be vital to recognizing the action. For example, actions like typing and clapping request the cooperation of two hands, which are distant from each other in a human skeleton graph. Multiple graph convolutional layers thus tend to be stacked together to increase receptive field, which brings in computational inefficiency and optimization difficulty. But there is still no guarantee that distant joints (e.g. two hands) can be well integrated. In this paper, we propose a context aware graph convolutional network (CA-GCN). Besides the computation of localized graph convolution, CA-GCN considers a context term for each vertex by integrating information of all other vertices. Long range dependencies among joints are thus naturally integrated in context information, which then eliminates the need of stacking multiple layers to enlarge receptive field and greatly simplifies the network. Moreover, we further propose an advanced CA-GCN, in which asymmetric relevance measurement and higher level representation are utilized to compute context information for more flexibility and better performance. Besides the joint features, our CA-GCN could also be extended to handle graphs with edge (limb) features. Extensive experiments on two real-world datasets demonstrate the importance of context information and the effectiveness of the proposed CA-GCN in skeleton based action recognition.

This paper studies the compression and acceleration of 3-dimensional convolutional neural networks (3D CNNs). To reduce the memory cost and computational complexity of deep neural networks, a number of algorithms have been explored by discovering redundant parameters in pre-trained networks. However, most of existing methods are designed for processing neural networks consisting of 2-dimensional convolution filters (i.e. image classification and detection) and cannot be straightforwardly applied for 3-dimensional filters (i.e. time series data). In this paper, we develop a novel approach for eliminating redundancy in the time dimensionality of 3D convolution filters by converting them into the frequency domain through a series of learned optimal transforms with extremely fewer parameters. Moreover, these transforms are forced to be orthogonal, and the calculation of feature maps can be accomplished in the frequency domain to achieve considerable speed-up rates. Experimental results on benchmark 3D CNN models and datasets demonstrate that the proposed Frequency Domain Compact 3D CNNs (FDC3D) can achieve the state-of-the-art performance, \eg a 2x speed-up ratio on the 3D-ResNet-18 without obviously affecting its accuracy.

Compared with cheap addition operation, multiplication operation is of much higher computation complexity. The widely-used convolutions in deep neural networks are exactly cross-correlation to measure the similarity between input feature and convolution filters, which involves massive multiplications between float values. In this paper, we present adder networks (AdderNets) to trade these massive multiplications in deep neural networks, especially convolutional neural networks (CNNs), for much cheaper additions to reduce computation costs. In AdderNets, we take the L1-norm distance between filters and input feature as the output response. The influence of this new similarity measure on the optimization of neural network have been thoroughly analyzed. To achieve a better performance, we develop a special back-propagation approach for AdderNets by investigating the full-precision gradient. We then propose an adaptive learning rate strategy to enhance the training procedure of AdderNets according to the magnitude of each neuron's gradient. As a result, the proposed AdderNets can achieve 74.9% Top-1 accuracy 91.7% Top-5 accuracy using ResNet-50 on the ImageNet dataset without any multiplication in convolutional layer. The codes are publicly available at: (https://github.com/huaweinoah/AdderNet).

Neural Architecture Search (NAS) has achieved great success in image classification task. Some recent works have managed to explore the automatic design of efficient backbone or feature fusion layer for object detection. However, these methods focus on searching only one certain component of object detector while leaving others manually designed. We identify the inconsistency between searched component and manually designed ones would withhold the detector of stronger performance. To this end, we propose a hierarchical trinity search framework to simultaneously discover efficient architectures for all components (i.e. backbone, neck, and head) of object detector in an end-to-end manner. In addition, we empirically reveal that different parts of the detector prefer different operators. Motivated by this, we employ a novel scheme to automatically screen different sub search spaces for different components so as to perform the end-to-end search for each component on the corresponding sub search space efficiently. Without bells and whistles, our searched architecture, namely Hit-Detector, achieves 41.4% mAP on COCO minival set with 27M parameters. Our implementation is available at https://github.com/ggjy/HitDet.pytorch.

This paper defines a positive and unlabeled classification problem for standard GANs, which then leads to a novel technique to stabilize the training of the discriminator in GANs. Traditionally, real data are taken as positive while generated data are negative. This positive-negative classification criterion was kept fixed all through the learning process of the discriminator without considering the gradually improved quality of generated data, even if they could be more realistic than real data at times. In contrast, it is more reasonable to treat the generated data as unlabeled, which could be positive or negative according to their quality. The discriminator is thus a classifier for this positive and unlabeled classification problem, and we derive a new Positive-Unlabeled GAN (PUGAN). We theoretically discuss the global optimality the proposed model will achieve and the equivalent optimization goal. Empirically, we find that PUGAN can achieve comparable or even better performance than those sophisticated discriminator stabilization methods.

Deploying convolutional neural networks (CNNs) on embedded devices is difficult due to the limited memory and computation resources. The redundancy in feature maps is an important characteristic of those successful CNNs, but has rarely been investigated in neural architecture design. This paper proposes a novel Ghost module to generate more feature maps from cheap operations. Based on a set of intrinsic feature maps, we apply a series of linear transformations with cheap cost to generate many ghost feature maps that could fully reveal information underlying intrinsic features. The proposed Ghost module can be taken as a plug-and-play component to upgrade existing convolutional neural networks. Ghost bottlenecks are designed to stack Ghost modules, and then the lightweight GhostNet can be easily established. Experiments conducted on benchmarks demonstrate that the proposed Ghost module is an impressive alternative of convolution layers in baseline models, and our GhostNet can achieve higher recognition performance (e.g. 75.7% top-1 accuracy) than MobileNetV3 with similar computational cost on the ImageNet ILSVRC-2012 classification dataset. Code is available at https://github.com/huawei-noah/ghostnet.

Reconstruction of high dynamic range image from a single low dynamic range image captured by a frame-based conventional camera, which suffers from over- or under-exposure, is an ill-posed problem. In contrast, recent neuromorphic cameras are able to record high dynamic range scenes in the form of an intensity map, with much lower spatial resolution, and without color. In this paper, we propose a neuromorphic camera guided high dynamic range imaging pipeline, and a network consisting of specially designed modules according to each step in the pipeline, which bridges the domain gaps on resolution, dynamic range, and color representation between two types of sensors and images. A hybrid camera system has been built to validate that the proposed method is able to reconstruct quantitatively and qualitatively high-quality high dynamic range images by successfully fusing the images and intensity maps for various real-world scenarios.

Neural architecture search (NAS) aims to automatically design deep neural networks of satisfactory performance. Wherein, architecture performance predictor is critical to efficiently value an intermediate neural architecture. But for the training of this predictor, a number of neural architectures and their corresponding real performance often have to be collected. In contrast with classical performance predictor optimized in a fully supervised way, this paper suggests a semi-supervised assessor of neural architectures. We employ an auto-encoder to discover meaningful representations of neural architectures. Taking each neural architecture as an individual instance in the search space, we construct a graph to capture their intrinsic similarities, where both labeled and unlabeled architectures are involved. A graph convolutional neural network is introduced to predict the performance of architectures based on the learned representations and their relation modeled by the graph. Extensive experimental results on the NAS-Benchmark-101 dataset demonstrated that our method is able to make a significant reduction on the required fully trained architectures for finding efficient architectures.

Searching techniques in most of existing neural architecture search (NAS) algorithms are mainly dominated by differentiable methods for the efficiency reason. In contrast, we develop an efficient continuous evolutionary approach for searching neural networks. Architectures in the population that share parameters within one SuperNet in the latest generation will be tuned over the training dataset with a few epochs. The searching in the next evolution generation will directly inherit both the SuperNet and the population, which accelerates the optimal network generation. The non-dominated sorting strategy is further applied to preserve only results on the Pareto front for accurately updating the SuperNet. Several neural networks with different model sizes and performances will be produced after the continuous search with only 0.4 GPU days. As a result, our framework provides a series of networks with the number of parameters ranging from 3.7M to 5.1M under mobile settings. These networks surpass those produced by the state-of-the-art methods on the benchmark ImageNet dataset.

We address the issue of having a limited number of annotations for stance classification in a new domain, by adapting out-of-domain classifiers with domain adaptation. Existing approaches often align different domains in a single, global feature space (or view), which may fail to fully capture the richness of the languages used for expressing stances, leading to reduced adaptability on stance data. In this paper, we identify two major types of stance expressions that are linguistically distinct, and we propose a tailored dual-view adaptation network (DAN) to adapt these expressions across domains. The proposed model first learns a separate view for domain transfer in each expression channel and then selects the best adapted parts of both views for optimal transfer. We find that the learned view features can be more easily aligned and more stance-discriminative in either or both views, leading to more transferable overall features after combining the views. Results from extensive experiments show that our method can enhance the state-of-the-art single-view methods in matching stance data across different domains, and that it consistently improves those methods on various adaptation tasks.

This paper formalizes the binarization operations over neural networks from a learning perspective. In contrast to classical hand crafted rules (\eg hard thresholding) to binarize full-precision neurons, we propose to learn a mapping from full-precision neurons to the target binary ones. Each individual weight entry will not be binarized independently. Instead, they are taken as a whole to accomplish the binarization, just as they work together in generating convolution features. To help the training of the binarization mapping, the full-precision neurons after taking sign operations is regarded as some auxiliary supervision signal, which is noisy but still has valuable guidance. An unbiased estimator is therefore introduced to mitigate the influence of the supervision noise. Experimental results on benchmark datasets indicate that the proposed binarization technique attains consistent improvements over baselines.

This paper searches for the optimal neural architecture by minimizing a proxy of validation loss. Existing neural architecture search (NAS) methods used to discover the optimal neural architecture that best fits the validation examples given the up-to-date network weights. However, back propagation with a number of validation examples could be time consuming, especially when it needs to be repeated many times in NAS. Though these intermediate validation results are invaluable, they would be wasted if we cannot use them to predict the future from the past. In this paper, we propose to approximate the validation loss landscape by learning a mapping from neural architectures to their corresponding validate losses. The optimal neural architecture thus can be easily identified as the minimum of this proxy validation loss landscape. A novel sampling strategy is further developed for an efficient approximation of the loss landscape. Theoretical analysis indicates that the validation loss estimator learnt with our sampling strategy can reach a lower error rate and a lower label complexity compared with a uniform sampling. Experimental results on benchmarks demonstrate that the architecture searched by the proposed algorithm can achieve a satisfactory accuracy with less time cost.

Image compression, as one of the fundamental low-level image processing tasks, is very essential for computer vision. Tremendous computing and storage resources can be preserved with a trivial amount of visual information. Conventional image compression methods tend to obtain compressed images by minimizing their appearance discrepancy with the corresponding original images, but pay little attention to their efficacy in downstream perception tasks, e.g., image recognition and object detection. Thus, some of compressed images could be recognized with bias. In contrast, this paper aims to produce compressed images by pursuing both appearance and perceptual consistency. Based on the encoder-decoder framework, we propose using a pre-trained CNN to extract features of the original and compressed images, and making them similar. Thus the compressed images are discernible to subsequent tasks, and we name our method as Discernible Image Compression (DIC). In addition, the maximum mean discrepancy (MMD) is employed to minimize the difference between feature distributions. The resulting compression network can generate images with high image quality and preserve the consistent perception in the feature domain, so that these images can be well recognized by pre-trained machine learning models. Experiments on benchmarks demonstrate that images compressed by using the proposed method can also be well recognized by subsequent visual recognition and detection models. For instance, the mAP value of compressed images by DIC is about 0.6% higher than that of using compressed images by conventional methods.

Neural architecture search (NAS) has demonstrated impressive performance in automatically designing high-performance neural networks. The power of deep neural networks is to be unleashed for analyzing a large volume of data (e.g. ImageNet), but the architecture search is often executed on another smaller dataset (e.g. CIFAR-10) to finish it in a feasible time. However, it is hard to guarantee that the optimal architecture derived on the proxy task could maintain its advantages on another more challenging dataset. This paper aims to improve the generalization of neural architectures via domain adaptation. We analyze the generalization bounds of the derived architecture and suggest its close relations with the validation error and the data distribution distance on both domains. These theoretical analyses lead to AdaptNAS, a novel and principled approach to adapt neural architectures between domains in NAS. Our experimental evaluation shows that only a small part of ImageNet will be sufficient for AdaptNAS to extend its architecture success to the entire ImageNet and outperform state-of-the-art comparison algorithms.

This paper proposes a reliable neural network pruning algorithm by setting up a scientific control. Existing pruning methods have developed various hypotheses to approximate the importance of filters to the network and then execute filter pruning accordingly. To increase the reliability of the results, we prefer to have a more rigorous research design by including a scientific control group as an essential part to minimize the effect of all factors except the association between the filter and expected network output. Acting as a control group, knockoff feature is generated to mimic the feature map produced by the network filter, but they are conditionally independent of the example label given the real feature map. We theoretically suggest that the knockoff condition can be approximately preserved given the information propagation of network layers. Besides the real feature map on an intermediate layer, the corresponding knockoff feature is brought in as another auxiliary input signal for the subsequent layers. Redundant filters can be discovered in the adversarial process of different features. Through experiments, we demonstrate the superiority of the proposed algorithm over state-of-the-art methods. For example, our method can reduce 57.8% parameters and 60.2% FLOPs of ResNet-101 with only 0.01% top-1 accuracy loss on ImageNet.

Adder Neural Networks (ANNs) which only contain additions bring us a new way of developing deep neural networks with low energy consumption. Unfortunately, there is an accuracy drop when replacing all convolution filters by adder filters. The main reason here is the optimization difficulty of ANNs using $\ell_1$-norm, in which the estimation of gradient in back propagation is inaccurate. In this paper, we present a novel method for further improving the performance of ANNs without increasing the trainable parameters via a progressive kernel based knowledge distillation (PKKD) method. A convolutional neural network (CNN) with the same architecture is simultaneously initialized and trained as a teacher network, features and weights of ANN and CNN will be transformed to a new space to eliminate the accuracy drop. The similarity is conducted in a higher-dimensional space to disentangle the difference of their distributions using a kernel based method. Finally, the desired ANN is learned based on the information from both the ground-truth and teacher, progressively. The effectiveness of the proposed method for learning ANN with higher performance is then well-verified on several benchmarks. For instance, the ANN-50 trained using the proposed PKKD method obtains a 76.8\% top-1 accuracy on ImageNet dataset, which is 0.6\% higher than that of the ResNet-50.

Quantized neural networks with low-bit weights and activations are attractive for developing AI accelerators. However, the quantization functions used in most conventional quantization methods are non-differentiable, which increases the optimization difficulty of quantized networks. Compared with full-precision parameters (\emph{i.e.}, 32-bit floating numbers), low-bit values are selected from a much smaller set. For example, there are only 16 possibilities in 4-bit space. Thus, we present to regard the discrete weights in an arbitrary quantized neural network as searchable variables, and utilize a differential method to search them accurately. In particular, each weight is represented as a probability distribution over the discrete value set. The probabilities are optimized during training and the values with the highest probability are selected to establish the desired quantized network. Experimental results on benchmarks demonstrate that the proposed method is able to produce quantized neural networks with higher performance over the state-of-the-arts on both image classification and super-resolution tasks.

A conventional camera often suffers from over- or under-exposure when recording a real-world scene with a very high dynamic range (HDR). In contrast, a modulo camera with a Markov random field (MRF) based unwrapping algorithm can theoretically accomplish unbounded dynamic range but shows degenerate performances when there are modulus-intensity ambiguity, strong local contrast, and color misalignment. In this paper, we reformulate the modulo image unwrapping problem into a series of binary labeling problems and propose a modulo edge-aware model, named as UnModNet, to iteratively estimate the binary rollover masks of the modulo image for unwrapping. Experimental results show that our approach can generate 12-bit HDR images from 8-bit modulo images reliably, and runs much faster than the previous MRF-based algorithm thanks to the GPU acceleration.

Video prediction refers to predicting and generating future video frames given a set of consecutive frames. Conventional video prediction methods usually criticize the discrepancy between the ground-truth and predictions frame by frame. As the prediction error accumulates recursively, these methods would easily become out of control and are often confined to the short-term horizon. In this paper, we introduce a retrospection process to rectify the prediction errors beyond criticizing the future prediction. The introduced retrospection process is designed to look back what have been learned from the past and rectify the prediction deficiencies. To this end, we build a retrospection network to reconstruct the past frames given the currently predicted frames. A retrospection loss is introduced to push the retrospection frames being consistent with the observed frames, so that the prediction error is alleviated. On the other hand, an auxiliary route is built by reversing the flow of time and executing a similar retrospection. These two routes interact with each other to boost the performance of retrospection network and enhance the understanding of dynamics across frames, especially for the long-term horizon. An adversarial loss is employed to generate more realistic results in both prediction and retrospection process. In addition, the proposed method can be used to extend many state-of-the-art video prediction methods. Extensive experiments on the natural video dataset demonstrate the advantage of introducing the retrospection process for long-term video prediction.

Conventional multi-view re-ranking methods usually perform asymmetrical matching between the region of interest (ROI) in the query image and the whole target image for similarity computation. Due to the inconsistency in the visual appearance, this practice tends to degrade the retrieval accuracy particularly when the image ROI, which is usually interpreted as the image objectness, accounts for a smaller region in the image. Since Privileged Information (PI), which can be viewed as the image prior, is able to characterize well the image objectness, we are aiming at leveraging PI for further improving the performance of multi-view re-ranking in this paper. Towards this end, we propose a discriminative multi-view re-ranking approach in which both the original global image visual contents and the local auxiliary PI features are simultaneously integrated into a unified training framework for generating the latent subspaces with sufficient discriminating power. For the on-the-fly re-ranking, since the multi-view PI features are unavailable, we only project the original multi-view image representations onto the latent subspace, and thus the re-ranking can be achieved by computing and sorting the distances from the multi-view embeddings to the separating hyperplane. Extensive experimental evaluations on the two public benchmarks, Oxford5k and Paris6k, reveal that our approach provides further performance boost for accurate image re-ranking, whilst the comparative study demonstrates the advantage of our method against other multi-view re-ranking methods.

Deep neural networks bring in impressive accuracy in various applications, but the success often relies on heavy network architectures. Taking well-trained heavy networks as teachers, classical teacher-student learning paradigm aims to learn a student network that is lightweight yet accurate. In this way, a portable student network with significantly fewer parameters can achieve considerable accuracy, which is comparable to that of a teacher network. However, beyond accuracy, the robustness of the learned student network against perturbation is also essential for practical uses. Existing teacher-student learning frameworks mainly focus on accuracy and compression ratios, but ignore the robustness. In this paper, we make the student network produce more confident predictions with the help of the teacher network, and analyze the lower bound of the perturbation that will destroy the confidence of the student network. Two important objectives regarding prediction scores and gradients of examples are developed to maximize this lower bound, to enhance the robustness of the student network without sacrificing the performance. Experiments on benchmark data sets demonstrate the efficiency of the proposed approach to learning robust student networks that have satisfying accuracy and compact sizes.

While in the learning using privileged information paradigm, privileged information may not be as informative as example features in the context of making accurate label predictions, it may be able to provide some effective comments (e.g., the values of the auxiliary function) like a human teacher on the efficacy of the learned model. In a departure from conventional static manipulations of privileged information within the support vector machine framework, this paper investigates iterative privileged learning within the context of gradient boosted decision trees (GBDTs). As the learned model evolves, the comments learned from privileged information to assess the model should also be actively upgraded instead of remaining static and passive. During the learning phase of the GBDT method, new DTs are discovered to enhance the performance of the model, and iteratively update the comments generated from the privileged information to accurately assess and coach the up-to-date model. The resulting objective function can be efficiently solved within the gradient boosting framework. Experimental results on real-world data sets demonstrate the benefits of studying privileged information in an iterative manner, as well as the effectiveness of the proposed algorithm.

Body joints, directly obtained from a pose estimation model, have proven effective for action recognition. Existing works focus on analyzing the dynamics of human joints. However, except joints, humans also explore motions of limbs for understanding actions. Given this observation, we investigate the dynamics of human limbs for skeleton-based action recognition. Specifically, we represent an edge in a graph of a human skeleton by integrating its spatial neighboring edges (for encoding the cooperation between different limbs) and its temporal neighboring edges (for achieving the consistency of movements in an action). Based on this new edge representation, we devise a graph edge convolutional neural network (CNN). Considering the complementarity between graph node convolution and edge convolution, we further construct two hybrid networks by introducing different shared intermediate layers to integrate graph node and edge CNNs. Our contributions are twofold, graph edge convolution and hybrid networks for integrating the proposed edge convolution and the conventional node convolution. Experimental results on the Kinetics and NTU-RGB+D data sets demonstrate that our graph edge convolution is effective at capturing the characteristics of actions and that our graph edge CNN significantly outperforms the existing state-of-the-art skeleton-based action recognition methods.

Generative Adversarial Networks (GANs) have demonstrated a strong ability to fit complex distributions since they were presented, especially in the field of generating natural images. Linear interpolation in the noise space produces a continuously changing in the image space, which is an impressive property of GANs. However, there is no special consideration on this property in the objective function of GANs or its derived models. This paper analyzes the perturbation on the input of the generator and its influence on the generated images. A smooth generator is then developed by investigating the tolerable input perturbation. We further integrate this smooth generator with a gradient penalized discriminator, and design smooth GAN that generates stable and high-quality images. Experiments on real-world image datasets demonstrate the necessity of studying smooth generator and the effectiveness of the proposed algorithm.

Geographic information systems’ (GIS) research is widely used within the social and physical sciences and plays a crucial role in the development and implementation by governments of economic, education, environment and transportation policy. While machine learning methods have been applied to GIS datasets, the uptake of powerful deep learning CNN methodologies has been limited in part due to challenges posed by the complex and often poorly structured nature of the data. In this paper, we demonstrate the utility of GCNNs for GIS analysis via a multi-graph hierarchical spatial-filter GCNN network model in the context of GIS systems to predict election outcomes using socio-economic features drawn from the 2016 Australian Census. We report a marked improvement in performance accuracy of Hierarchical GCNNs over benchmark generalised linear models and standard GCNNs, especially in semi-supervised tasks. These results indicate the widespread potential for GIS-GCNN research methods to enrich socio-economic GIS analysis, aiding the social sciences and policy development.

We identify agreement and disagreement between utterances that express stances towards a topic of discussion.Existing methods focus mainly on conversational settings, where dialogic features are used for (dis)agreement inference.We extend this scope and seek to detect stance (dis)agreement in a broader setting, where independent stance-bearing utterances, which prevail in many stance corpora and real-world scenarios, are compared.To cope with such non-dialogic utterances, we find that the reasons uttered to back up a specific stance can help predict stance (dis)agreements.We propose a reason comparing network (RCN) to leverage reason information for stance comparison.Empirical results on a well-known stance corpus show that our method can discover useful reason information, enabling it to outperform several baselines in stance (dis)agreement detection.

A recent trojan attack on deep neural network (DNN) models is one insidious variant of data poisoning attacks. Trojan attacks exploit an effective backdoor created in a DNN model by leveraging the difficulty in interpretability of the learned model to misclassify any inputs signed with the attacker's chosen trojan trigger. Since the trojan trigger is a secret guarded and exploited by the attacker, detecting such trojan inputs is a challenge, especially at run-time when models are in active operation. This work builds STRong Intentional Perturbation (STRIP) based run-time trojan attack detection system and focuses on vision system. We intentionally perturb the incoming input, for instance by superimposing various image patterns, and observe the randomness of predicted classes for perturbed inputs from a given deployed model---malicious or benign. A low entropy in predicted classes violates the input-dependence property of a benign model and implies the presence of a malicious input---a characteristic of a trojaned input. The high efficacy of our method is validated through case studies on three popular and contrasting datasets: MNIST, CIFAR10 and GTSRB. We achieve an overall false acceptance rate (FAR) of less than 1%, given a preset false rejection rate (FRR) of 1%, for different types of triggers. Using CIFAR10 and GTSRB, we have empirically achieved result of 0% for both FRR and FAR. We have also evaluated STRIP robustness against a number of trojan attack variants and adaptive attacks.

We introduce a self-supervised learning method that focuses on beneficial properties of representation and their abilities in generalizing to real-world tasks. The method incorporates rotation invariance into the feature learning framework, one of many good and well-studied properties of visual representation, which is rarely appreciated or exploited by previous deep convolutional neural network based self-supervised representation learning methods. Specifically, our model learns a split representation that contains both rotation related and unrelated parts. We train neural networks by jointly predicting image rotations and discriminating individual instances. In particular, our model decouples the rotation discrimination from instance discrimination, which allows us to improve the rotation prediction by mitigating the influence of rotation label noise, as well as discriminate instances without regard to image rotations. The resulting feature has a better generalization ability for more various tasks. Experimental results show that our model outperforms current state-of-the-art methods on standard self-supervised feature learning benchmarks.

The image, question (combined with the history for de-referencing), and the corresponding answer are three vital components of visual dialog. Classical visual dialog systems integrate the image, question, and history to search for or generate the best matched answer, and so, this approach significantly ignores the role of the answer. In this paper, we devise a novel image-question-answer synergistic network to value the role of the answer for precise visual dialog. We extend the traditional one-stage solution to a two-stage solution. In the first stage, candidate answers are coarsely scored according to their relevance to the image and question pair. Afterward, in the second stage, answers with high probability of being correct are re-ranked by synergizing with image and question. On the Visual Dialog v1.0 dataset, the proposed synergistic network boosts the discriminative visual dialog model to achieve a new state-of-the-art of 57.88% normalized discounted cumulative gain. A generative visual dialog model equipped with the proposed technique also shows promising improvements.

Watermark is one of the most fundamental approaches for avoiding potential copyright infringement activities. However, whether its introduction would effect the understanding of deep learning models remains unstudied. In this work, we propose a visible adversarial attack method that transforms and places a provided watermark on the target image to interfere the classification result from an Inception V3 model, which is pretrained on ImageNet. Specifically, the watermark is adjusted iteratively on location, transparency, color, angle and size which are determined by only 9 parameters. We define two types of attack to better simulate the watermark approaches in reality, respectively the watermark is constrained in either transparency or size. Experiments show that the generated adversarial samples are not only capable of fooling the Inception V3 model with high success rates, but also transferable to other models with high confidence, such as the Rekognition developed by Amazon.

Learning portable neural networks is very essential for computer vision for the purpose that pre-trained heavy deep models can be well applied on edge devices such as mobile phones and micro sensors. Most existing deep neural network compression and speed-up methods are very effective for training compact deep models, when we can directly access the training dataset. However, training data for the given deep network are often unavailable due to some practice problems (\eg privacy, legal issue, and transmission), and the architecture of the given network are also unknown except some interfaces. To this end, we propose a novel framework for training efficient deep neural networks by exploiting generative adversarial networks (GANs). To be specific, the pre-trained teacher networks are regarded as a fixed discriminator and the generator is utilized for derivating training samples which can obtain the maximum response on the discriminator. Then, an efficient network with smaller model size and computational complexity is trained using the generated data and the teacher network, simultaneously. Efficient student networks learned using the proposed Data-Free Learning (DFL) method achieve 92.22% and 74.47% accuracies without any training data on the CIFAR-10 and CIFAR-100 datasets, respectively. Meanwhile, our student network obtains an 80.56% accuracy on the CelebA benchmark.

We present an information-theoretically motivated constraint for self-supervised representation learning from multiple related domains. In contrast to previous self-supervised learning methods, our approach learns from multiple domains, which has the benefit of decreasing the build-in bias of individual domain, as well as leveraging information and allowing knowledge transfer across multiple domains. The proposed mutual information constraints encourage neural network to extract common invariant information across domains and to preserve peculiar information of each domain simultaneously. We adopt tractable upper and lower bounds of mutual information to make the proposed constraints solvable. The learned representation is more unbiased and robust toward the input images. Extensive experimental results on both multi-domain and large-scale datasets demonstrate the necessity and advantage of multi-domain self-supervised learning with mutual information constraints. Representations learned in our framework on state-of-the-art methods achieve improved performance than those learned on a single domain.

Generative adversarial networks (GANs) have been successfully used for considerable computer vision tasks, especially the image-to-image translation. However, generators in these networks are of complicated architectures with large number of parameters and huge computational complexities. Existing methods are mainly designed for compressing and speeding-up deep neural networks in the classification task, and cannot be directly applied on GANs for image translation, due to their different objectives and training procedures. To this end, we develop a novel co-evolutionary approach for reducing their memory usage and FLOPs simultaneously. In practice, generators for two image domains are encoded as two populations and synergistically optimized for investigating the most important convolution filters iteratively. Fitness of each individual is calculated using the number of parameters, a discriminator-aware regularization, and the cycle consistency. Extensive experiments conducted on benchmark datasets demonstrate the effectiveness of the proposed method for obtaining compact and effective generators.

We consider two less-emphasized temporal properties of video: 1. Temporal cues are fine-grained; 2. Temporal modeling needs reasoning. To tackle both problems at once, we exploit approximated bilinear modules (ABMs) for temporal modeling. There are two main points making the modules effective: two-layer MLPs can be seen as a constraint approximation of bilinear operations, thus can be used to construct deep ABMs in existing CNNs while reusing pretrained parameters; frame features can be divided into static and dynamic parts because of visual repetition in adjacent frames, which enables temporal modeling to be more efficient. Multiple ABM variants and implementations are investigated, from high performance to high efficiency. Specifically, we show how two-layer subnets in CNNs can be converted to temporal bilinear modules by adding an auxiliary-branch. Besides, we introduce snippet sampling and shifting inference to boost sparse-frame video classification performance. Extensive ablation studies are conducted to show the effectiveness of proposed techniques. Our models can outperform most state-of-the-art methods on Something-Something v1 and v2 datasets without Kinetics pretraining, and are also competitive on other YouTube-like action recognition datasets. Our code is available on https://github.com/zhuxinqimac/abm-pytorch.

De novo molecular design and generation are frequently prescribed in the field of chemistry and biology, for it plays a critical role in maintaining the prosperity of the chemical industry and benefiting the drug discovery. In recent years, reinforcement learning-based methods leverage graphs to represent molecules and generate molecules as a decision making process. However, this vanilla graph representation may neglect the intrinsic context information with molecules and limits the generation performance accordingly. In this paper, we propose to augment the original graph states with the SMILES context vectors. As a result, SMILES representations are easily processed by a simple language model such that the general semantic features of a molecule can be extracted; and the graph representations perform better in handling the topology relationship of each atom. Moreover, we propose a framework that combines supervised learning and reinforcement learning algorithm to take a solid consideration of these two heterogeneous state representations of a molecule, which can fuse the information from both of them and extract more comprehensive features so that more sophisticated decisions can be made by the policy network. Our model also introduces two attention mechanisms, i.e., action-attention, and graph-attention, to further improve the performance. We conduct our experiments on a practical dataset, ZINC, and the experiment results demonstrate that our framework can outperform other baselines in the learning performance of molecule generation and chemical property optimization.

This paper aims to build efficient convolutional neural networks using a set of Lego filters. Many successful building blocks, e.g., inception and residual modules, have been designed to refresh state-of-the-art records of CNNs on visual recognition tasks. Beyond these high-level modules, we suggest that an ordinary filter in the neural network can be upgraded to a sophisticated module as well. Filter modules are established by assembling a shared set of Lego filters that are often of much lower dimensions. Weights in Lego filters and binary masks to stack Lego filters for these filter modules can be simultaneously optimized in an end-to-end manner as usual. Inspired by network engineering, we develop a split-transform-merge strategy for an efficient convolution by exploiting intermediate Lego feature maps. The compression and acceleration achieved by Lego Networks using the proposed Lego filters have been theoretically discussed. Experimental results on benchmark datasets and deep models demonstrate the advantages of the proposed Lego filters and their potential real-world applications on mobile devices.

Network pruning is widely applied to deep CNN models due to their heavy computation costs and achieves high performance by keeping important weights while removing the redundancy. Pruning redundant weights directly may hurt global information flow, which suggests that an efficient sparse network should take graph properties into account. Thus, instead of paying more attention to preserving important weight, we focus on the pruned architecture itself. We propose to use graph entropy as the measurement, which shows useful properties to craft high-quality neural graphs and enables us to propose efficient algorithm to construct them as the initial network architecture. Our algorithm can be easily implemented and deployed to different popular CNN models and achieve better trade-offs.

Deep recurrent neural networks have achieved impressive success in forecasting human motion with a sequence to sequence architecture. However, forecasting in longer time horizons often leads to implausible human poses or converges to mean poses, because of error accumulation and difficulties in keeping track of longerterm information. To address these challenges, we propose to retrospect human dynamics with attention. A retrospection module is designed upon RNN to regularly retrospect past frames and correct mistakes in time. This significantly improves the memory of RNN and provides sufficient information for the decoder networks to generate longer term prediction. Moreover, we present a spatial attention module to explore and exploit cooperation among joints in performing a particular motion. Residual connections are also included to guarantee the performance of short term prediction. We evaluate the proposed algorithm on the largest and most challenging Human 3.6M dataset in the field. Experimental results demonstrate the necessity of investigating motion prediction in a self audit manner and the effectiveness of the proposed algorithm in both short term and long term predictions.

This paper expands the strength of deep convolutional neural networks (CNNs) to the pedestrian attribute recognition problem by devising a novel attribute aware pooling algorithm. Existing vanilla CNNs cannot be straightforwardly applied to handle multi-attribute data because of the larger label space as well as the attribute entanglement and correlations. We tackle these challenges that hampers the development of CNNs for multi-attribute classification by fully exploiting the correlation between different attributes. The multi-branch architecture is adopted for fucusing on attributes at different regions. Besides the prediction based on each branch itself, context information of each branch are employed for decision as well. The attribute aware pooling is developed to integrate both kinds of information. Therefore, attributes which are indistinct or tangled with others can be accurately recognized by exploiting the context information. Experiments on benchmark datasets demonstrate that the proposed pooling method appropriately explores and exploits the correlations between attributes for the pedestrian attribute recognition.

Exploring deep convolutional neural networks of high efficiency and low memory usage is very essential for a wide variety of machine learning tasks. Most of existing approaches used to accelerate deep models by manipulating parameters or filters without data, e.g., pruning and decomposition. In contrast, we study this problem from a different perspective by respecting the difference between data. An instance-wise feature pruning is developed by identifying informative features for different instances. Specifically, by investigating a feature decay regularization, we expect intermediate feature maps of each instance in deep neural networks to be sparse while preserving the overall network performance. During online inference, subtle features of input images extracted by intermediate layers of a well-trained neural network can be eliminated to accelerate the subsequent calculations. We further take coefficient of variation as a measure to select the layers that are appropriate for acceleration. Extensive experiments conducted on benchmark datasets and networks demonstrate the effectiveness of the proposed method.

Bad training data would challenge the learning model from understanding the underlying data-generating scheme, which then increases the difficulty in achieving satisfactory performance on unseen test data. We suppose the real data distribution lies in a distribution set supported by the empirical distribution of bad data. A worst-case formulation can be developed over this distribution set, and then be interpreted as a generation task in an adversarial manner. The connections and differences between GANs and our framework have been thoroughly discussed. We further theoretically show the influence of this generation task on learning from bad data and reveal its connection with a data-dependent regularization. Given different distance measures (\eg, Wasserstein distance or JS divergence) of distributions, we can derive different objective functions for the problem. Experimental results on different kinds of bad training data demonstrate the necessity and effectiveness of the proposed method.

Many attempts have been done to extend the great success of convolutional neural networks (CNNs) achieved on high-end GPU servers to portable devices such as smart phones. Providing compression and acceleration service of deep learning models on the cloud is therefore of significance and is attractive for end users. However, existing network compression and acceleration approaches usually fine-tuning the svelte model by requesting the entire original training data (e.g. ImageNet), which could be more cumbersome than the network itself and cannot be easily uploaded to the cloud. In this paper, we present a novel positive-unlabeled (PU) setting for addressing this problem. In practice, only a small portion of the original training set is required as positive examples and more useful training examples can be obtained from the massive unlabeled data on the cloud through a PU classifier with an attention based multi-scale feature extractor. We further introduce a robust knowledge distillation (RKD) scheme to deal with the class imbalance problem of these newly augmented training examples. The superiority of the proposed method is verified through experiments conducted on the benchmark models and datasets. We can use only 8% of uniformly selected data from the ImageNet to obtain an efficient model with comparable performance to the baseline ResNet-34.

Deep convolutional neural networks (CNNs) are successfully used in a number of applications. However, their storage and computational requirements have largely prevented their widespread use on mobile devices. Here we present a series of approaches for compressing and speeding up CNNs in the frequency domain, which focuses not only on smaller weights but on all the weights and their underlying connections. By treating convolution filters as images, we decompose their representations in the frequency domain as common parts (i.e., cluster centers) shared by other similar filters and their individual private parts (i.e., individual residuals). A large number of low-energy frequency coefficients in both parts can be discarded to produce high compression without significantly compression romising accuracy. Furthermore, we explore a data-driven method for removing redundancies in both spatial and frequency domains, which allows us to discard more useless weights by keeping similar accuracies. After obtaining the optimal sparse CNN in the frequency domain, we relax the computational burden of convolution operations in CNNs by linearly combining the convolution responses of discrete cosine transform (DCT) bases. The compression and speed-up ratios of the proposed algorithm are thoroughly analyzed and evaluated on benchmark image datasets to demonstrate its superiority over state-of-the-art methods.

Generative adversarial networks (GANs) have been effective for learning generative models for real-world data. However, accompanied with the generative tasks becoming more and more challenging, existing GANs (GAN and its variants) tend to suffer from different training problems such as instability and mode collapse. In this paper, we propose a novel GAN framework called evolutionary GANs (E-GANs) for stable GAN training and improved generative performance. Unlike existing GANs, which employ a predefined adversarial objective function alternately training a generator and a discriminator, we evolve a population of generators to play the adversarial game with the discriminator. Different adversarial training objectives are employed as mutation operations and each individual (i.e., generator candidature) are updated based on these mutations. Then, we devise an evaluation mechanism to measure the quality and diversity of generated samples, such that only well-performing generator(s) are preserved and used for further training. In this way, E-GAN overcomes the limitations of an individual adversarial training objective and always preserves the well-performing offspring, contributing to progress in, and the success of GANs. Experiments on several datasets demonstrate that E-GAN achieves convincing generative performance and reduces the training problems inherent in existing GANs.

Style transfer describes the rendering of an image's semantic content as different artistic styles. Recently, generative adversarial networks (GANs) have emerged as an effective approach in style transfer by adversarially training the generator to synthesize convincing counterfeits. However, traditional GAN suffers from the mode collapse issue, resulting in unstable training and making style transfer quality difficult to guarantee. In addition, the GAN generator is only compatible with one style, so a series of GANs must be trained to provide users with choices to transfer more than one kind of style. In this paper, we focus on tackling these challenges and limitations to improve style transfer. We propose adversarial gated networks (Gated-GAN) to transfer multiple styles in a single model. The generative networks have three modules: an encoder, a gated transformer, and a decoder. Different styles can be achieved by passing input images through different branches of the gated transformer. To stabilize training, the encoder and decoder are combined as an auto-encoder to reconstruct the input images. The discriminative networks are used to distinguish whether the input image is a stylized or genuine image. An auxiliary classifier is used to recognize the style categories of transferred images, thereby helping the generative networks generate images in multiple styles. In addition, Gated-GAN makes it possible to explore a new style by investigating styles learned from artists or genres. Our extensive experiments demonstrate the stability and effectiveness of the proposed model for multi-style transfer.

In multi-view re-ranking, multiple heterogeneous visual features are usually projected onto a low-dimensional subspace, and thus the resulting latent representation can be used for the subsequent similarity-based ranking. Albeit effective, this standard mechanism underplays the intrinsic structure underlying the latent subspace and does not take into account the substantial noise in the original spaces. In this paper, we propose a robust multi-view image re-ranking strategy. Due to the dramatic variability in image visual appearance, it is necessary to uncover the shared components underlying those query-related instances that are visually unlike for improving the re-ranking accuracy. Consequently, it is reasonable to assume the latent subspace enjoys the low-rank property and thus the subspace recovery can be achieved via the low-rank modeling accordingly. In addition, since the real-world data are usually partially contaminated, we employ `2;1-norm based sparsity constraint to appropriately model the sample-specific mapping noise for enhancing the model robustness. In order to produce discriminative representations, we encode a similarity preserving term in our multi-view embedding framework. As a result, the sample separability is maximally maintained in the latent subspace with sufficient discriminative power. The extensive evaluations on public landmark benchmarks demonstrate the efficacy and superiority of the proposed method.

Recommendation systems have become ubiquitous in online shopping in recent decades due to their power in reducing excessive choices of customers and industries. Recent collaborative filtering methods based on the deep neural network are studied and introduce promising results due to their power in learning hidden representations for users and items. However, it has revealed its vulnerabilities under malicious user attacks. With the knowledge of a collaborative filtering algorithm and its parameters, the performance of this recommendation system can be easily downgraded. Unfortunately, this problem is not addressed well, and the study on defending recommendation systems is insufficient. In this paper, we aim to improve the robustness of recommendation systems based on two concepts - stage-wise hints training and randomness. To protect a target model, we introduce noise layers in the training of a target model to increase its resistance to adversarial perturbations. To reduce the noise layers' influence on model performance, we introduce intermediate layer outputs as hints from a teacher model to regularize the intermediate layers of a student target model. We consider white box attacks under which attackers have the knowledge of the target model. The generalizability and robustness properties of our method have been analytically inspected in experiments and discussions, and the computational cost is comparable to training a standard neural network-based collaborative filtering model. Through our investigation, the proposed defensive method can reduce the success rate of malicious user attacks and keep the prediction accuracy comparable to standard neural recommendation systems.

Humans and animals learn much better when the examples are not randomly presented but organized in a meaningful order which illustrates gradually more concepts, and gradually more complex ones. Inspired by this curriculum learning mechanism, we propose a reinforced multi-label image classification approach imitating human behavior to label image from easy to complex. This approach allows a reinforcement learning agent to sequentially predict labels by fully exploiting image feature and previously predicted labels. The agent discovers the optimal policies through maximizing the long-term reward which reflects prediction accuracies. Experimental results on PASCAL VOC2007 and 2012 demonstrate the necessity of reinforcement multi-label learning and the algorithm’s effectiveness in real-world multi-label image classification tasks.

Effective methods for learning deep neural networks with fewer parameters are urgently required, since storage and computations of heavy neural networks have largely prevented their widespread use on mobile devices. Compared with algorithms which directly remove weights or filters for obtaining considerable compression and speed-up ratios, training thin deep networks exploiting the student-teacher learning paradigm is more flexible. However, it is very hard to determine which formulation is optimal to measure the information inherited from teacher networks. To overcome this challenge, we utilize the generative adversarial network (GAN) to learn the student network. In practice, the generator is exactly the student network with extremely less parameters and the discriminator is used as a teaching assistant for distinguishing features extracted from student and teacher networks. By simultaneously optimizing the generator and the discriminator, the resulting student network can produce features of input data with the similar distribution as that of features of the teacher network. Extensive experimental results on benchmark datasets demonstrate that the proposed method is capable of learning well-performed portable networks, which is superior to the state-of-the-art methods.

In this paper we study a new learning problem defined as "Single-Teacher Multi-Student" (STMS) problem, which investigates how to learn a series of student (simple and specific) models from a single teacher (complex and universal) model. Taking the multiclass and binary classification for example, we focus on learning multiple binary classifiers from a single multiclass classifier, where each of binary classifier is responsible for a certain class. This actually derives from some realistic problems, such as identifying the suspect based on a comprehensive face recognition system. By treating the already-trained multiclass classifier as the teacher, and multiple binary classifiers as the students, we propose a gated support vector machine (gSVM) as a solution. A series of gSVMs are learned with the help of single teacher multiclass classifier. The teacher's help is two-fold; first, the teacher's score provides the gated values for students' decision; second, the teacher can guide the students to accommodate training examples with different difficulty degrees. Extensive experiments on real datasets validate its effectiveness.

Online dictionary learning has received intensive attention in signal processing field with streaming or dynamic data. Different from classical online dictionary learning methods that treat all atoms equally, in this paper, we present a novel online dictionary learning with a confidence parameter introduced on each of atoms. The confidence indicates the scale of the update of atoms during online learning; frequently estimated atoms are thus supposed to be updated less aggressively than those of low usage frequency. This updating mechanism is beneficial for learning with dirty examples. As a result, the outliers would be prevented from influencing much on the frequently used atoms that have been well estimated through the clean examples. In detail, we employ variance of the atoms to measure the confidence on the quality of the learned atoms. And frequently-used atoms are supposed to have smaller variances (i.e. more confidence) than those of low usage frequency. Our algorithm consists of two main parts, namely, confidence-weighted sparse coding and dictionary update with confidence fine-tuning, each of which can be solved efficiently by our designed optimization methods. In addition, due to the introduced confidence, our algorithm does not have to depend on the widely-used ℓ1norm in order to realize the robustness against the outliers, which is usually computationally expensive in the training cost. Experimental results on synthetic and benchmark datasets demonstrate the imposed confidence on atoms in the dictionary can improve the performance of the learned dictionary. The quality of the obtained dictionary is comparable to that of state-of-the-art robust online dictionary learning methods, however, our method is much faster than those ℓ1norm based ones.

Investigating correlation between example features and example labels is essential to solve classification problems. However, identification and calculation of the correlation between features and labels can be rather difficult for high-dimensional multi-label data. Both feature embedding and label embedding have been developed to tackle this challenge, and a shared subspace for both labels and features are usually learned by existing embedding methods to simultaneously reduce dimensionality of features and labels. In contrast, this paper suggests to learn separated subspaces for features and labels by maximizing the independence between components in each subspace and maximizing the correlation between these two subspaces. The learned independent label components indicates fundamental combinations of labels in multi-label datasets, which thus helps to reveals the correlation between labels. On the other hand, the learned independent feature components lead to a compact representation of example features. The connections between the proposed algorithm and existing embedding methods have been discussed. Experimental results on real-world multi-label datasets demonstrate the necessity of exploring independence components from multi-label data and the effectiveness of the proposed algorithm.

In practice, the circumstance that training and test data are clean is not always satisfied. The performance of existing methods in the learning using privileged information (LUPI) paradigm may be seriously challenged, due to the lack of clear strategies to address potential noises in the data. This paper proposes a novel Robust SVM+ (RSVM+) algorithm based on a rigorous theoretical analysis. Under the SVM+ framework in the LUPI paradigm, we study the lower bound of perturbations of both example feature data and privileged feature data, which will mislead the model to make wrong decisions. By maximizing the lower bound, tolerance of the learned model over perturbations will be increased. Accordingly, a novel regularization function is introduced to upgrade a variant form of SVM+. The objective function of RSVM+ is transformed into a quadratic programming problem, which can be efficiently optimized using off-the-shelf solvers. Experiments on real-world datasets demonstrate the necessity of studying robust SVM+ and the effectiveness of the proposed algorithm.

Compressing convolutional neural networks (CNNs) is essential for transferring the success of CNNs to a wide variety of applications to mobile devices. In contrast to directly recognizing subtle weights or filters as redundant in a given CNN, this paper presents an evolutionary method to automatically eliminate redundant convolution filters. We represent each compressed network as a binary individual of specific fitness. Then, the population is upgraded at each evolutionary iteration using genetic operations. As a result, an extremely compact CNN is generated using the fittest individual, which has the original network structure and can be directly deployed in any off-the-shelf deep learning libraries. In this approach, either large or small convolution filters can be redundant, and filters in the compressed network are more distinct. In addition, since the number of filters in each convolutional layer is reduced, the number of filter channels and the size of feature maps are also decreased, naturally improving both the compression and speed-up ratios. Experiments on benchmark deep CNN models suggest the superiority of the proposed algorithm over the state-of-the-art compression methods, e.g. combined with the parameter refining approach, we can reduce the storage requirement and the floating-point multiplications of ResNet-50 by a factor of 14.64x and 5.19x, respectively, without affecting its accuracy.

This paper introduces versatile filters to construct efficient convolutional neural network. Considering the demands of efficient deep learning techniques running on cost-effective hardware, a number of methods have been developed to learn compact neural networks. Most of these works aim to slim down filters in different ways, e.g., investigating small, sparse or binarized filters. In contrast, we treat filters from an additive perspective. A series of secondary filters can be derived from a primary filter. These secondary filters all inherit in the primary filter without occupying more storage, but once been unfolded in computation they could significantly enhance the capability of the filter by integrating information extracted from different receptive fields. Besides spatial versatile filters, we additionally investigate versatile filters from the channel perspective. The new techniques are general to upgrade filters in existing CNNs. Experimental results on benchmark datasets and neural networks demonstrate that CNNs constructed with our versatile filters are able to achieve comparable accuracy as that of original filters, but require less memory and FLOPs.

This paper studies the collaborative budget allocation problem in which users are not isolated in the collaborative consumption of goods or services when available goods or services are limited. Different from existing methods that treat each user independently, we investigate the geometric properties of user’s consumption or preference on services, and design a matrix completion framework on the simplex. In this framework, an item’s allocation vector indicating how available services are allocated to users is estimated by the combination of user profiles as basis points on the simplex. Instead of using Euclidean distance directly, we specify a Riemannian distance on the simplex or project histogram data on simplex to Euclidean space. To intensify our model’s stability, we relax the exact recovery constraint to make a robust collaborative prediction. The resulting objective function is then efficiently optimized by a Riemannian conjugate gradient method on the simplex. Experiments on real-world data sets demonstrate our model’s competitiveness versus other collaborative budget prediction methods. Comparisons of different distance metrics for histogram data are shown and discussed.Note to Practitioners—This paper was motivated by the collaborative budget allocation (CBA) problem in which either a user has limited ratings that can be distributed among items, or a service has limited availability to all users in the sharing economy system. Existing approaches rarely consider this phenomenon in their collaborative prediction modeling, which implies the geometric constraints on a user’s rating vector. This paper suggests a matrix factorization framework to address this problem by learning a group of a user profile as basis points that can be combined to recover other users’ rating vectors. Considering the noise in observed ratings in real applications, we enhance the robustness of collaborative prediction by relaxing the recovery constraint. We then propose to specify the distance metric on the simplex from two perspectives, and give the optimization approach. In the future research, we will address the CBA in more complex settings, for example, some items can be substitutes or complementary for each other, in this way, how items can be better distributed among users is an interesting problem.

Blind source separation (BSS) aims to discover the underlying source signals from a set of linear mixture signals without any prior information of the mixing system, which is a fundamental problem in signal and image processing field. Most of the state-of-the-art algorithms have independently handled the decompositions of mixture signals. In this paper, we propose a new algorithm named multi-task sparse model to solve the BSS problem. Source signals are characterized via sparse techniques. Meanwhile, we regard the decomposition of each mixture signal as a task and employ the idea of multi-task learning to discover connections between tasks for the accuracy improvement of the source signal separation. Theoretical analyses on the optimization convergence and sample complexity of the proposed algorithm are provided. Experimental results based on extensive synthetic and real-world data demonstrate the necessity of exploiting connections between mixture signals and the effectiveness of the proposed algorithm.

Feature selection is beneficial for improving the performance of general machine learning tasks by extracting an informative subset from the high-dimensional features. Conventional feature selection methods usually ignore the class imbalance problem, thus the selected features will be biased towards the majority class. Considering that F-measure is a more reasonable performance measure than accuracy for imbalanced data, this paper presents an effective feature selection algorithm that explores the class imbalance issue by optimizing F-measures. Since F-measure optimization can be decomposed into a series of cost-sensitive classification problems, we investigate the cost-sensitive feature selection by generating and assigning different costs to each class with rigorous theory guidance. After solving a series of cost-sensitive feature selection problems, features corresponding to the best F-measure will be selected. In this way, the selected features will fully represent the properties of all classes. Experimental results on popular benchmarks and challenging real-world data sets demonstrate the significance of cost-sensitive feature selection for the imbalanced data setting and validate the effectiveness of the proposed method.

In this paper, we propose Perceptual Adversarial Networks (PAN) for image-to-image transformations. Different from existing application driven algorithms, PAN provides a generic framework of learning to map from input images to desired images (Fig. 1), such as a rainy image to its de-rained counterpart, object edges to photos, semantic labels to a scenes image, etc. The proposed PAN consists of two feed-forward convolutional neural networks (CNNs): the image transformation network T and the discriminative network D. Besides the generative adversarial loss widely used in GANs, we propose the perceptual adversarial loss, which undergoes an adversarial training process between the image transformation network T and the hidden layers of the discriminative network D. The hidden layers and the output of the discriminative network D are upgraded to constantly and automatically discover the discrepancy between the transformed image and the corresponding ground-truth, while the image transformation network T is trained to minimize the discrepancy explored by the discriminative network D. Through integrating the generative adversarial loss and the perceptual adversarial loss, D and T can be trained alternately to solve image-to-image transformation tasks. Experiments evaluated on several image-to-image transformation tasks (e.g., image de-raining, image inpainting, etc) demonstrate the effectiveness of the proposed PAN and its advantages over many existing works.

Feature selection aims to select a small subset from the high-dimensional features which can lead to better learning performance, lower computational complexity, and better model readability. The class imbalance problem has been neglected by traditional feature selection methods, therefore the selected features will be biased towards the majority classes. Because of the superiority of F-measure to accuracy for imbalanced data, we propose to use F-measure as the performance measure for feature selection algorithms. As a pseudo-linear function, the optimization of F-measure can be achieved by minimizing the total costs. In this paper, we present a novel cost-sensitive feature selection (CSFS) method which optimizes F-measure instead of accuracy to take class imbalance issue into account. The features will be selected according to optimal F-measure classifier after solving a series of cost-sensitive feature selection sub-problems. The features selected by our method will fully represent the characteristics of not only majority classes, but also minority classes. Extensive experimental results conducted on synthetic, multi-class and multi-label datasets validate the efficiency and significance of our feature selection method.

Discovering robust low-rank data representations is important in many real-world problems. Traditional robust principal component analysis (RPCA) assumes that the observed data are corrupted by some sparse noise (e.g., Laplacian noise) and utilizes the l1-norm to separate out the noisy compo- nent. Nevertheless, as well as simple Gaussian or Laplacian noise, noise in real-world data is often more complex, and thus the l1 and l2-norms are insufficient for noise charac- terization. This paper presents a more flexible approach to modeling complex noise by investigating their properties in the frequency domain. Although elements of a noise matrix are chaotic in the spatial domain, the absolute values of its alternative coefficients in the frequency domain are constant w.r.t. their variance. Based on this observation, a new robust PCA algorithm is formulated by simultaneously discovering the low-rank and noisy components. Extensive experiments on synthetic data and video background subtraction demon- strate that FRPCA is effective for handles complex noise.

Convolutional neural networks (CNNs) have shown extraordinary performance in a number of applications, but they are usually of heavy design for the accuracy reason. Beyond compressing the filters in CNNs, this paper focuses on the redundancy in the feature maps derived from the large number of filters in a layer. We propose to extract intrinsic representation of the feature maps and preserve the discriminability of the features. Circulant matrix is employed to formulate the feature map transformation, which only requires O(dlog d) computation complexity to embed a d-dimensional feature map. The filter is then re-configured to establish the mapping from original input to the new compact feature map, and the resulting network can preserve intrinsic information of the original network with significantly fewer parameters, which not only decreases the online memory for launching CNN but also accelerates the computation speed. Experiments on benchmark image datasets demonstrate the superiority of the proposed algorithm over state-of-the-art methods.

Collaborative filtering plays a crucial role in reducing excessive information in online consuming by suggesting products to customers that fulfil their potential interests. Observing that a user's review comments on purchases are often in companion with ratings, recent works exploit the review texts in representing user or item factors and have achieved prominent performance. Although effectiveness of reviews has been verified, one major defect of existing works is that reviews are used in justifying the learning of either user or item factors without noticing that each review associates a pair of user and item concurrently. To better explore the value of review comments, this paper presents the privileged matrix factorization method that utilize reviews in the learning of both user and item factors. By mapping review texts into the privileged feature space, a learned privileged function compensates the discrepancies between predicted ratings and groundtruth values rating-wisely. Thus by minimizing discrepancies and prediction errors, our method harnesses the information present in the review comments for the learning of both user and item factors. Experiments on five real datasets testify the effectiveness of the proposed method.

This paper studies the collaborative rating allocation problem, in which each user has limited ratings on all items. These users are termed ``energy limited''. Different from existing methods which treat each rating independently, we investigate the geometric properties of a user's rating vector, and design a matrix completion method on the simplex. In this method, a user's rating vector is estimated by the combination of user profiles as basis points on the simplex. Instead of using Euclidean metric, a non-linear pull-back distance measurement from the sphere is adopted since it can depict the geometric constraints on each user's rating vector. The resulting objective function is then efficiently optimized by a Riemannian conjugate gradient method on the simplex. Experiments on real-world data sets demonstrate our model's competitiveness versus other collaborative rating prediction methods.

Supporting vector machine (SVM) is the most frequently used classifier for machine learning tasks. However, its training time could become cumbersome when the size of training data is very large. Thus, many kinds of representative subsets are chosen from the original dataset to reduce the training complexity. In this paper, we propose to choose the representative points which are noted as anchors obtained from non-negative matrix factorization (NMF) in a divide-and-conquer framework, and then use the anchors to train an approximate SVM. Our theoretical analysis shows that the solving the DCA-SVM can yield an approximate solution close to the primal SVM. Experimental results on multiple datasets demonstrate that our DCA-SVM is faster than the state-of-the-art algorithms without notably decreasing the accuracy of classification results.

In this paper, we propose a principled Tag Disentangled Generative Adversarial Networks (TD-GAN) for re-rendering new images for the object of interest from a single image of it by specifying multiple scene properties (such as viewpoint, illumination, expression, etc.). The whole framework consists of a disentangling network, a generative network, a tag mapping net, and a discriminative network, which are trained jointly based on a given set of images that are completely/partially tagged (i.e., supervised/semi-supervised setting). Given an input image, the disentangling network extracts disentangled and interpretable representations, which are then used to generate images by the generative network. In order to boost the quality of disentangled representations, the tag mapping net is integrated to explore the consistency between the image and its tags. Furthermore, the discriminative network is introduced to implement the adversarial training strategy for generating more realistic images. Experiments on two challenging datasets demonstrate the state-of-the-art performance of the proposed framework in the problem of interest.

The positive and unlabeled (PU) learning problem focuses on learning a classifier from positive and unlabeled data. Some methods have been developed to solve the PU learning problem. However, they are often limited in practical applications, since only binary classes are involved and cannot easily be adapted to multi-class data. Here we propose a one-step method that directly enables multi-class model to be trained using the given input multi-class data and that predicts the label based on the model decision. Specifically, we construct different convex loss functions for labeled and unlabeled data to learn a discriminant function F. The theoretical analysis on the generalization error bound shows that it is no worse than k√k times of the fully supervised multi-class classification methods when the size of the data in k classes is of the same order. Finally, our experimental results demonstrate the significance and effectiveness of the proposed algorithm in synthetic and real-world datasets.

This paper presents privileged multi-label learning (PrML) to explore and exploit the relationship between labels in multi-label learning problems. We suggest that for each individual label, it cannot only be implicitly connected with other labels via the low-rank constraint over label predictors, but also its performance on examples can receive the explicit comments from other labels together acting as an Oracle teacher. We generate privileged label feature for each example and its individual label, and then integrate it into the framework of low-rank based multi-label learning. The proposed algorithm can therefore comprehensively explore and exploit label relationships by inheriting all the merits of privileged information and low-rank constraints. We show that PrML can be efficiently solved by dual coordinate descent algorithm using iterative optimization strategy with cheap updates. Experiments on benchmark datasets show that through privileged label features, the performance can be significantly improved and PrML is superior to several competing methods in most cases.

Training thin deep networks following the student-teacher learning paradigm has received intensive attention because of its excellent performance. However, to the best of our knowledge, most existing work mainly considers one single teacher network. In practice, a student may access multiple teachers, and multiple teacher networks together provide comprehensive guidance that is beneficial for training the student network. In this paper, we present a method to train a thin deep network by incorporating multiple teacher networks not only in output layer by averaging the softened outputs (dark knowledge) from different networks, but also in the intermediate layers by imposing a constraint about the dissimilarity among examples. We suggest that the relative dissimilarity between intermediate representations of different examples serves as a more flexible and appropriate guidance from teacher networks. Then triplets are utilized to encourage the consistence of these relative dissimilarity relationships between the student network and teacher networks. Moreover, we leverage a voting strategy to unify multiple relative dissimilarity information provided by multiple teacher networks, which realizes their incorporation in the intermediate layers. Extensive experimental results demonstrated that our method is capable of generating a well-performed student network, with the classification accuracy comparable or even superior to all teacher networks, yet having much fewer parameters and being much faster in running.

Given an unreliable visual patterns and insufficient query information, content-based image retrieval is often suboptimal and requires image re-ranking using auxiliary information. In this paper, we propose a discriminative multi-view interactive image re-ranking (DMINTIR), which integrates user relevance feedback capturing users' intentions and multiple features that sufficiently describe the images. In DMINTIR, heterogeneous property features are incorporated in the multi-view learning scheme to exploit their complementarities. In addition, a discriminatively learned weight vector is obtained to reassign updated scores and target images for re-ranking. Compared with other multi-view learning techniques, our scheme not only generates a compact representation in the latent space from the redundant multi-view features but also maximally preserves the discriminative information in feature encoding by the large-margin principle. Furthermore, the generalization error bound of the proposed algorithm is theoretically analyzed and shown to be improved by the interactions between the latent space and discriminant function learning. Experimental results on two benchmark data sets demonstrate that our approach boosts baseline retrieval quality and is competitive with the other state-of-the-art re-ranking strategies.

Here we study the extreme visual recovery problem, in which over 90% of pixel values in a given image are missing. Existing low rank-based algorithms are only effective for recovering data with at most 90% missing values. Thus, we exploit visual data's smoothness property to help solve this challenging extreme visual recovery problem. Based on the discrete cosine transform (DCT), we propose a novel DCT regularizer that involves all pixels and produces smooth estimations in any view. Our theoretical analysis shows that the total variation regularizer, which only achieves local smoothness, is a special case of the proposed DCT regularizer. We also develop a new visual recovery algorithm by minimizing the DCT regularizer and nuclear norm to achieve a more visually pleasing estimation. Experimental results on a benchmark image data set demonstrate that the proposed approach is superior to the state-of-the-art methods in terms of peak signal-to-noise ratio and structural similarity.

Tail labels in the multi-label learning problem undermine the low-rank assumption. Nevertheless, this problem has rarely been investigated. In addition to using the low-rank structure to depict label correlations, this paper explores and exploits an additional sparse component to handle tail labels behaving as outliers, in order to make the classical low-rank principle in multi-label learning valid. The divide-and-conquer optimization technique is employed to increase the scalability of the proposed algorithm while theoretically guaranteeing its performance. A theoretical analysis of the generalizability of the proposed algorithm suggests that it can be improved by the low-rank and sparse decomposition given tail labels. Experimental results on real-world data demonstrate the significance of investigating tail labels and the effectiveness of the proposed algorithm.

Deep convolutional neural networks (CNNs) are successfully used in a number of applications. However, their storage and computational requirements have largely prevented their widespread use on mobile devices. Here we present an effective CNN compression approach in the frequency domain, which focuses not only on smaller weights but on all the weights and their underlying connections. By treating convolutional filters as images, we decompose their representations in the frequency domain as common parts (i.e., cluster centers) shared by other similar filters and their individual private parts (i.e., individual residuals). A large number of low-energy frequency coefficients in both parts can be discarded to produce high compression without significantly compromising accuracy. We relax the computational burden of convolution operations in CNNs by linearly combining the convolution responses of discrete cosine transform (DCT) bases. The compression and speed-up ratios of the proposed algorithm are thoroughly analyzed and evaluated on benchmark image datasets to demonstrate its superiority over state-of-the-art methods.

We analyze the local Rademacher complexity of empirical risk minimization-based multi-label learning algorithms, and in doing so propose a new algorithm for multi-label learning. Rather than using the trace norm to regularize the multi-label predictor, we instead minimize the tail sum of the singular values of the predictor in multi-label learning. Benefiting from the use of the local Rademacher complexity, our algorithm, therefore, has a sharper generalization error bound. Compared with methods that minimize over all singular values, concentrating on the tail singular values results in better recovery of the low-rank structure of the multi-label predictor, which plays an important role in exploiting label correlations. We propose a new conditional singular value thresholding algorithm to solve the resulting objective function. Moreover, a variance control strategy is employed to reduce the variance of variables in optimization. Empirical studies on real-world data sets validate our theoretical results and demonstrate the effectiveness of the proposed algorithm for multi-label learning.

In multi-label learning, an example is represented by a descriptive feature associated with several labels. Simply considering labels as independent or correlated is crude; it would be beneficial to define and exploit the causality between multiple labels. For example, an image label 'lake' implies the label 'water', but not vice versa. Since the original features are a disorderly mixture of the properties originating from different labels, it is intuitive to factorize these raw features to clearly represent each individual label and its causality relationship.Following the large-margin principle, we propose an effective approach to discover the causal features of multiple labels, thus revealing the causality between labels from the perspective of feature. We show theoretically that the proposed approach is a tight approximation of the empirical multi-label classification error, and the causality revealed strengthens the consistency of the algorithm. Extensive experimentations using synthetic and real-world data demonstrate that the proposed algorithm effectively discovers label causality, generates causal features, and improves multi-label learning.

It is practical to assume that an individual view is unlikely to be sufficient for effective multi-view learning. Therefore, integration of multi-view information is both valuable and necessary. In this paper, we propose the Multi-view Intact Space Learning (MISL) algorithm, which integrates the encoded complementary information in multiple views to discover a latent intact representation of the data. Even though each view on its own is insufficient, we show theoretically that by combing multiple views we can obtain abundant information for latent intact space learning. Employing the Cauchy loss (a technique used in statistical learning) as the error measurement strengthens robustness to outliers. We propose a new definition of multi-view stability and then derive the generalization error bound based on multi-view stability and Rademacher complexity, and show that the complementarity between multiple views is beneficial for the stability and generalization. MISL is efficiently optimized using a novel Iteratively Reweight Residuals (IRR) technique, whose convergence is theoretically analyzed. Experiments on synthetic data and real-world datasets demonstrate that MISL is an effective and promising algorithm for practical applications.

Face images captured in unconstrained environments usually contain significant pose variation, which dramatically degrades the performance of algorithms designed to recognize frontal faces. This paper proposes a novel face identification framework capable of handling the full range of pose variations within ±90° of yaw. The proposed framework first transforms the original pose-invariant face recognition problem into a partial frontal face recognition problem. A robust patch-based face representation scheme is then developed to represent the synthesized partial frontal faces. For each patch, a transformation dictionary is learnt under the proposed multi-task learning scheme. The transformation dictionary transforms the features of different poses into a discriminative subspace. Finally, face matching is performed at patch level rather than at the holistic level. Extensive and systematic experimentation on FERET, CMU-PIE, and Multi-PIE databases shows that the proposed method consistently outperforms single-task-based baselines as well as state-of-the-art methods for the pose problem. We further extend the proposed algorithm for the unconstrained face verification problem and achieve top-level performance on the challenging LFW data set.

One underlying assumption of the conventional multi-view learning algorithms is that all examples can be successfully observed on all the views. However, due to various failures or faults in collecting and pre-processing the data on different views, we are more likely to be faced with an incomplete-view setting, where an example could be missing its representation on one view (i.e., missing view) or could be only partially observed on that view (i.e., missing variables). Low-rank assumption used to be effective for recovering the random missing variables of features, but it is disabled by concentrated missing variables and has no effect on missing views. This paper suggests that the key to handling the incomplete-view problem is to exploit the connections between multiple views, enabling the incomplete views to be restored with the help of the complete views. We propose an effective algorithm to accomplish multi-view learning with incomplete views by assuming that different views are generated from a shared subspace. To handle the large-scale problem and obtain fast convergence, we investigate a successive over-relaxation method to solve the objective function. Convergence of the optimization technique is theoretically analyzed. The experimental results on toy data and real-world data sets suggest that studying the incomplete-view problem in multi-view learning is significant and that the proposed algorithm can effectively handle the incomplete views in different applications.

This paper studies dimensionality reduction in a weakly supervised setting, in which the preference relationship between examples is indicated by weak cues. A novel framework is proposed that integrates two aspects of the large margin principle (angle and distance), which simultaneously encourage angle consistency between preference pairs and maximize the distance between examples in preference pairs. Two specific algorithms are developed: an alternating direction method to learn a linear transformation matrix and a gradient boosting technique to optimize a non-linear transformation directly in the function space. Theoretical analysis demonstrates that the proposed large margin optimization criteria can strengthen and improve the robustness and generalization performance of preference learning algorithms on the obtained low-dimensional subspace. Experimental results on real-world datasets demonstrate the significance of studying dimensionality reduction in the weakly supervised setting and the effectiveness of the proposed framework.

In this paper, we extend the theory of the information bottleneck (IB) to learning from examples represented by multi-view features. We formulate the problem as one of encoding a communication system with multiple senders, each of which represents one view of the data. Based on the precise components filtered out from multiple information sources through a "bottleneck", a margin maximization approach is then used to strengthen the discrimination of the encoder by improving the code distance within the frame of coding theory. The resulting algorithm therefore inherits all the merits of the IB principle and coding theory. It has two distinct advantages over existing algorithms, namely, that our method finds a tradeoff between the accuracy and complexity of the multi-view model, and that the encoded multi-view data retains sufficient discrimination for classification. We also derive the robustness and generalization error bound of the proposed algorithm, and reveal the specific properties of multi-view learning. First, the complementarity of multi-view features guarantees the robustness of the algorithm. Second, the consensus of multi-view features reduces the empirical Rademacher complexity of the objective function, enhances the accuracy of the solution, and improves the generalization error bound of the algorithm. The resulting objective function is solved efficiently using the alternating direction method. Experimental results on annotation, classification and recognition tasks demonstrate that the proposed algorithm is promising for practical applications.

Images are usually associated with multiple labels and comprised of multiple views, due to each image containing several objects (e.g. a pedestrian, bicycle and tree) and multiple visual features (e.g. color, texture and shape). Currently available tools tend to use either labels or features for classification, but both are necessary to describe the image properly. There have been recent successes in using vector-valued functions, which construct matrix-valued kernels, to explore the multi-label structure in the output space. This has motivated us to develop multi-view vector-valued manifold regularization (MV$^3$MR) in order to integrate multiple features. MV$^3$MR exploits the complementary properties of different features, and discovers the intrinsic local geometry of the compact support shared by different features, under the theme of manifold regularization. We validate the effectiveness of the proposed MV$^3$MR methodology for image classification by conducting extensive experiments on two challenge datasets, PASCAL VOC' 07 and MIR Flickr.

In computer vision, image datasets used for classification are naturally associated with multiple labels and comprised of multiple views, because each image may contain several objects (e.g., pedestrian, bicycle, and tree) and is properly characterized by multiple visual features (e.g., color, texture, and shape). Currently, available tools ignore either the label relationship or the view complementarily. Motivated by the success of the vector-valued function that constructs matrix-valued kernels to explore the multilabel structure in the output space, we introduce multiview vector-valued manifold regularization (MV(3)MR) to integrate multiple features. MV(3)MR exploits the complementary property of different features and discovers the intrinsic local geometry of the compact support shared by different features under the theme of manifold regularization. We conduct extensive experiments on two challenging, but popular, datasets, PASCAL VOC' 07 and MIR Flickr, and validate the effectiveness of the proposed MV(3)MR for image classification.

Reranking is one of the commonly used schemes to improve the initial ranking performance for content based image retrieval (CBIR). The state-of-the-art reranking methods for CBIR are mainly based on spatial verification and global feature. To mine the complementary properties of different reranking strategies, we combine features representing images from different perspectives with RankSVM to obtain a reranking model to refine the initial ranking list. Besides, compared with traditional spatial verification based methods which measure image similarity only with single inlier's statistical properties, we bind close inlier visual words together to mine more geometric information from images. Through organizing inliers into sequence and computing the relative positions among inliers, we define an efficient similarity measurement with the order consistency between inlier sequences. Experimental results on both Oxford and imageNet datasets demonstrate that our proposed reranking method is effective and promising.