We propose a lightweight, modular framework for hallucination detection in Retrieval-Augmented Generation (RAG) systems, addressing the critical challenge where logical dependencies span across fragmented retrieval results.Through graph-based semantic evidence aggregation, which captures the implicit logical structure by clustering semantically coherent segments across retrieved documents via betweenness centrality, our approach enables small NLI models to handle multihop reasoning without task-specific training.We present two deployment configurations: a resource-efficient variant (≈0.5B parameters) achieving 82.4% accuracy on HotPotQA-Derived at 85 ms latency, outperforming all sub-1B baselines by over 30%; and a higheraccuracy variant (≈1.5B parameters) reaching 85.6%, surpassing 11B TrueTeacher while being 7× smaller and 1.7× faster.Experiments with six NLI discriminator models show consistent gains of +6.7%-+29.9%,confirming that graph-based evidence aggregation is NLIagnostic and the primary performance driver.We also contribute HotPotQA-Derived, a new multi-hop hallucination benchmark preserving separate retrieved documents for systematic evaluation.

Text-conditioned diffusion models have revolutionized the field of controllable real image editing, enabling high-fidelity and precise image manipulation. Recent methods target specific editing tasks, using internal representations from reconstruction to ensure consistency. Although effective for single tasks, they fail to balance precision and consistency across diverse image editing tasks. In this work, we propose a novel inference-time real-image editing framework that enables executing multiple editing tasks by tuning editing operators. Our key insight is to treat real image editing as a multi-objective optimization problem, optimizing editing operators for a Pareto optimal solution that balances editing accuracy and consistency at each denoising iteration. Additionally, we design a benchmark for operator-guided real-image editing that covers various local and global editing tasks. Extensive experimental evaluations demonstrate the method’s effectiveness in executing precise edits while preserving image fidelity across all tasks, thereby establishing it as the new state-of-the-art.

Attributing to the strong reasoning capability, behavior models in artificial intelligence for games play a crucial role in creating gaming experiences. For further enhancing the reasoning capability of behavior models, we propose a novel Switch, Reason, and Revise (SRR) framework, which integrates them with Large Language Model (LLM). The SRR framework contains three core components. The component of Dual-Track Experiential Reasoning fully utilizes the agent experiences for reasoning. The component of Block-Retrieval-Augmented Thoughts adaptively determines the granularity of information retrieval for external sources. The component of Self-Reliant Thinking System Switch increases the reasoning speed by model switching and performs the model switching automatically upon the LLM. Together, these three components can strengthen the agent reasoning capability in complex tasks. Experimental results in the Pokémon battle environment demonstrate the effectiveness and efficiency superiority of SRR over the rival methods. Furthermore, we conduct an exploratory study to reveal the potential of the SRR-empowered agent for guiding new players in Pokémon battle games.

Multiagent reinforcement learning (MARL) methods have good application performances and prospects in cooperative tasks. To improve the capability of agent policy learning in new scenarios, some methods transfer the learned policy knowledge to new scenarios. However, most methods only focus on the knowledge transfer of individual agent policies, neglecting the credit assignment among agents in cooperative tasks, which results in a transfer bias of cooperative policies. In this paper, we propose a novel method, transferable policy generation and credit assignment (TPGCA) network for cooperative MARL. TPGCA can transfer the entire MARL model by the constructed transferable$Q$-value network and mixing network. Specifically, in TPGCA, to enhance the effectivity and transferability of agent policies, we design the correspondence network between observations and actions (COA) on the basis of transformer and gated recurrent unit (GRU). To implement the reliable credit assignment and diminish the transfer bias, we devise the role-based joint$Q$-value decomposition network (RVD) that can evaluate the contributions of agents from different observation perspectives. Experimental results in various micro-management scenarios on StarCraft multiagent challenge (SMAC) and multiagent particle environment (MPE) sufficiently demonstrate the effectiveness and transferability of TPGCA.

Brain tumor segmentation is often based on multiple magnetic resonance imaging (MRI). However, in clinical practice, certain modalities of MRI may be missing, which presents a more difficult scenario. To cope with this challenge, Knowledge Distillation, Domain Adaption, and Shared Latent Space have emerged as commonly promising strategies. However, recent efforts to address the missing modality problem in brain tumor segmentation typically overlook the modality gaps and thus fail to learn important invariant feature representations across different modalities. Such drawback consequently leads to limited performance for missing modality models. To ameliorate these problems, pre-trained models are used in natural visual segmentation tasks to minimize the gaps. However, promising pre-trained models are difficult to obtain in the brain tumor segmentation task due to the lack of sufficient data. Along this line, in this paper, we propose a novel paradigm that aligns latent features of involved modalities to a well-defined distribution anchor as the substitution of the pre-trained model. As a major contribution, we prove that our novel training paradigm ensures a tight evidence lower bound, thus theoretically certifying its effectiveness. Extensive experiments on different backbones validate that the proposed paradigm can enable invariant feature representations and produce models with narrowed modality gaps. Models with our alignment paradigm show their superior performance on both BraTS2018, BraTS2020 and Brain Metastasis datasets.

Recognizing oracle bone scripts plays an important role in Chinese archaeology and philology. However, a significant challenge remains because of the scarcity of oracle character images. To overcome this issue, we propose Diff-Oracle, a novel multi-modal conditional diffusion model that generates a diverse range of controllable oracle characters by inputting random combinations of references. Given the challenge of accurately describing oracle character styles using natural language, Diff-Oracle departs from traditional diffusion models that rely primarily on text prompts by introducing a style encoder. This encoder extracts style prompts from existing oracle character images, where style details are converted into a text embedding format via a pre-trained language-vision model. Additionally, given the lack of explicit content information for oracle characters, ensuring that generated characters accurately represent the intended glyphs is challenging. Therefore, we pre-generate pixel-level paired oracle character images (i.e., style and content images) by an image-to-image translation model, providing content information for the generation process. Meanwhile, Diff-Oracle integrates a content encoder designed to capture specific content details from content reference images. Extensive experiments on Oracle-241 and OBC306 datasets demonstrate that Diff-Oracle significantly outperforms existing generative methods in image quality and diversity. Moreover, Diff-Oracle substantially benefits downstream recognition tasks, outperforming all existing state-of-the-art methods by a large margin. In particular, on the challenging OBC306 dataset, Diff-Oracle achieves a 7.70% accuracy gain in the zero-shot setting and reaches 84.62% accuracy for unseen oracle characters, setting a new benchmark for oracle character recognition. The code is available at https://github.com/JJJingLi/Diff-Oracle .

Solving tabular math word problems (TMWPs) has become a critical role in evaluating the mathematical reasoning ability of large language models (LLMs), where large-scale TMWP samples are commonly required for fine-tuning. Since the collection of high-quality TMWP datasets is costly and time-consuming, recent research has concentrated on automatic TMWP generation. However, current generated samples usually suffer from issues of either correctness or diversity. In this paper, we propose a Template-driven LLM-paraphrased (TeLL) framework for generating high-quality TMWP samples with diverse backgrounds and accurate tables, questions, answers, and solutions. To this end, we first extract templates from existing real samples to generate initial problems, ensuring correctness. Then, we adopt an LLM to extend templates and paraphrase problems, obtaining diverse TMWP samples. Furthermore, we find the reasoning annotation is important for solving TMWPs. Therefore, we propose to enrich each solution with illustrative reasoning steps. Through the proposed framework, we construct a high-quality dataset TabMWP-TeLL by adhering to the question types in the TabMWP dataset, and we conduct extensive experiments on a variety of LLMs to demonstrate the effectiveness of TabMWP-TeLL in improving TMWP-solving performance.

With the outstanding capabilities of Large Language Models (LLMs), solving math word problems (MWP) has greatly progressed, achieving higher performance on several benchmark datasets. However, it is more challenging to solve plane geometry problems (PGPs) due to the necessity of understanding, reasoning and computation on two modality data including both geometry diagrams and textual questions, where Multi-Modal Large Language Models (MLLMs) have not been extensively explored. Previous works simply regarded a plane geometry problem as multi-modal QA task, which ignored the importance of explicit parsing geometric elements from problems. To tackle this limitation, we propose to solve plane Geometry problems by Neural-Symbolic reasoning with MLLMs (GNS). We first leverage an MLLM to understand PGPs through knowledge prediction and symbolic parsing, next perform mathematical reasoning to obtain solutions, last adopt a symbolic solver to compute answers. Correspondingly, we introduce the largest PGPs dataset GNS-260K with multiple annotations including symbolic parsing, understanding, reasoning and computation. In experiments, our Phi3-Vision-based MLLM wins the first place on the PGPs solving task of MathVista benchmark, outperforming GPT-4o, Gemini Ultra and other much larger MLLMs. While LLaVA-13B-based MLLM markedly exceeded other close-source and open-source MLLMs on the MathVerse benchmark and also achieved the new SOTA on GeoQA dataset.

Marvelous advances have been exhibited in recent document tampering localization (DTL) systems. However, confronted with corrupted tampered document images, their vulnerability is fatal in real-world scenarios. While robustness against adversarial attack has been extensively studied by adversarial training (AT), the robustness on natural corruptions remains under-explored for DTL. In this paper, to overcome forensic dependency, we propose the adversarial forensic regularization (AFR) based on min-max optimization to improve robustness. Specifically, we adopt mutual information (MI) to represent forensic dependency between two random variable over tampered and authentic pixels spaces, where the MI can be approximated by Jensen-Shannon-Divergence (JSD) with empirical sampling. To further enable a trade-off between predictive representations in clean tampered document pixels and robust ones in corrupted pixels, an additional regularization term is formulated with divergence between clean and perturbed pixels distribution (DDR). Following min-max optimization framework, our method can also work well against adversarial attacks. To evaluate our proposed method, we collect a dataset (i.e., TSorie-CRP) for evaluating robustness against natural corruptions in real scenarios. Extensive experiments demonstrate the effectiveness of our method against natural corruptions. Without any surprise, our method also achieves good performance against adversarial attack on DTL benchmark datasets.

Tabular anomaly detection under the one-class classification setting poses a significant challenge, as it involves accurately conceptualizing "normal" derived exclusively from a single category to discern anomalies from normal data variations. Capturing the intrinsic correlation among attributes within normal samples presents one promising method for learning the concept. To do so, the most recent effort relies on a learnable mask strategy with a reconstruction task. However, this wisdom may suffer from the risk of producing uniform masks, i.e., essentially nothing is masked, leading to less effective correlation learning. To address this issue, we presume that attributes related to others in normal samples can be divided into two non-overlapping and correlated subsets, defined as CorrSets, to capture the intrinsic correlation effectively. Accordingly, we introduce an innovative method that disentangles CorrSets from normal tabular data. To our knowledge, this is a pioneering effort to apply the concept of disentanglement for one-class anomaly detection on tabular data. Extensive experiments on 20 tabular datasets show that our method substantially outperforms the state-of-the-art methods and leads to an average performance improvement of 6.1% on AUC-PR and 2.1% on AUC-ROC.

Magnetic resonance imaging (MRI), with modalities including T1, T2, T1ce, and Flair, providing complementary information critical for sub-region analysis, is widely used for brain tumor diagnosis. However, clinical practice often suffers from varying degrees of incompleteness of necessary modalities due to reasons such as susceptibility to artifacts. It significantly impairs segmentation model performance. Given the limited available modalities at hand, existing approaches attempt to project them into a shared latent space. However, they ignore decomposing the modality-shared and modality-specific information and failed to construct the relationship among different modalities. Such deficiency limits the effectiveness of the segmentation performance, particularly at the time when the amount of data in each modality is different. In this paper, we propose the plug-and-play Koopman Multi-modality Decomposition (KMD) module, leveraging the Koopman Invariant Subspace to disentangle modality-common and modality-specific information. It is capable of constructing modality relationships that minimize bias toward modalities across various modality-incomplete scenarios. More importantly, it can be integrated into several existing backbones feasibility. Through theoretical deductions and extensive empirical experiences on the BraTS2018 and BraTS2020 datasets, we have sufficiently demonstrated the effectiveness of the proposed KMD to promote generalization performance. Code is available at: https://github.com/T-Y-Liu/KMD.

Point cloud anomaly detection under the anomaly-free setting poses significant challenges as it requires accurately capturing the features of 3D normal data to identify deviations indicative of anomalies. Current efforts focus on devising reconstruction tasks, such as acquiring normal data representations by restoring normal samples from altered, pseudo-anomalous counterparts. Our findings reveal that distributing attention equally across normal and pseudo-anomalous data tends to dilute the model’s focus on anomalous deviations. The challenge is further compounded by the inherently disordered and sparse nature of 3D point cloud data. In response to those predicaments, we introduce an innovative approach that emphasizes learning point offsets, targeting more informative pseudo-abnormal points, thus fostering more effective distillation of normal data representations. We also have crafted an augmentation technique that is steered by normal vectors, facilitating the creation of credible pseudo anomalies that enhance the efficiency of the training process. Our comprehensive experimental evaluation on the Anomaly-ShapeNet and Real3DAD datasets evidences that our proposed method outperforms existing state-of-the-art approaches, achieving an average enhancement of 9.0% and 1.4% in the AUC-ROC detection metric across these datasets, respectively. Code is available at https://github.com/yjnanan/PO3AD.

3D semantic segmentation plays a fundamental and crucial role to understand 3D scenes. While contemporary state-of-the-art techniques predominantly concentrate on elevating the overall performance of 3D semantic segmentation based on general metrics (e.g. mIoU, mAcc, and oAcc), they unfortunately leave the exploration of challenging regions for segmentation mostly neglected. In this paper, we revisit 3D semantic segmentation through a more granular lens, shedding light on subtle complexities that are typically overshadowed by broader performance metrics. Concretely, we have delineated 3D semantic segmentation errors into four comprehensive categories as well as corresponding evaluation metrics tailored to each. Building upon this categorical framework, we introduce an innovative 3D semantic segmentation network called BFANet that incorporates detailed analysis of semantic boundary features. First, we design the boundary-semantic module to decouple point cloud features into semantic and boundary features, and fuse their query queue to enhance semantic features with attention. Second, we introduce a more concise and accelerated boundary pseudo-label calculation algorithm, which is 3.9 times faster than the state-of-the-art, offering compatibility with data augmentation and enabling efficient computation in training. Extensive experiments on benchmark data indicate the superiority of our BFANet model, confirming the significance of emphasizing the four uniquely designed metrics. Code is available at https://github.com/weiguangzhao/BFANet.

Xiaoqiang Kang, Shengen Wu, Zimu Wang, Yilin Liu, Xiaobo Jin, Kaizhu Huang, Wei Wang, Yutao Yue, Xiaowei Huang, Qiufeng Wang. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.

Zero-shot Learning (ZSL) aims to enable classifiers to identify unseen classes. This is typically achieved by generating visual features for unseen classes based on learned visual-semantic correlations from seen classes. However, most current generative approaches heavily rely on having a sufficient number of samples from seen classes. Our study reveals that a scarcity of seen class samples results in a marked decrease in performance across many generative ZSL techniques. We argue, quantify, and empirically demonstrate that this decline is largely attributable to spurious visual-semantic correlations. To address this issue, we introduce ZeroDiff, an innovative generative framework for ZSL that incorporates diffusion mechanisms and contrastive representations to enhance visual-semantic correlations. ZeroDiff comprises three key components: (1) Diffusion augmentation, which naturally transforms limited data into an expanded set of noised data to mitigate generative model overfitting; (2) Supervised-contrastive (SC)-based representations that dynamically characterize each limited sample to support visual feature generation; and (3) Multiple feature discriminators employing a Wasserstein-distance-based mutual learning approach, evaluating generated features from various perspectives, including pre-defined semantics, SC-based representations, and the diffusion process. Extensive experiments on three popular ZSL benchmarks demonstrate that ZeroDiff not only achieves significant improvements over existing ZSL methods but also maintains robust performance even with scarce training data. Our codes are available at https://github.com/FouriYe/ZeroDiff_ICLR25.

Exceptional mathematical reasoning ability is one of the key features that demonstrate the power of large language models (LLMs). How to comprehensively define and evaluate the mathematical abilities of LLMs, and even reflect the user experience in real-world scenarios, has emerged as a critical issue. Current benchmarks predominantly concentrate on problem-solving capabilities, presenting a substantial risk of model overfitting and fails to accurately measure the genuine mathematical reasoning abilities. In this paper, we argue that if a model really understands a problem, it should be robustly and readily applied across a diverse array of tasks. To this end, we introduce MathCheck, a well-designed checklist for testing task generalization and reasoning robustness, as well as an automatic tool to generate checklists efficiently. MathCheck includes multiple mathematical reasoning tasks and robustness tests to facilitate a comprehensive evaluation of both mathematical reasoning ability and behavior testing. Utilizing MathCheck, we develop MathCheck-GSM and MathCheck-GEO to assess mathematical textual reasoning and multi-modal reasoning capabilities, respectively, serving as upgraded versions of benchmarks including GSM8k, GeoQA, UniGeo, and Geometry3K. We adopt MathCheck-GSM and MathCheck-GEO to evaluate over 26 LLMs and 17 multi-modal LLMs, assessing their comprehensive mathematical reasoning abilities. Our results demonstrate that while frontier LLMs like GPT-4o continue to excel in various abilities on the checklist, many other model families exhibit a significant decline. Further experiments indicate that, compared to traditional math benchmarks, MathCheck better reflects true mathematical abilities and represents mathematical intelligence more linearly, thereby supporting our design. Using MathCheck, we can also efficiently conduct informative behavior analysis to deeply investigate models. Finally, we show that our proposed checklist paradigm can easily extend to other reasoning tasks for their comprehensive evaluation.

While weakly supervised multi-view face reconstruction (MVR) is garnering increased attention, one critical issue still remains open: how to effectively interact and fuse multiple image information to reconstruct high-precision 3D models. In this regard, we propose a novel pipeline called Deep Fusion MVR (DF-MVR) to explore the feature correspondences between multi-view images and reconstruct high-precision 3D faces. Specifically, we present a novel multi-view feature fusion backbone that utilizes face masks to align features from multiple encoders and integrates one multi-layer attention mechanism to enhance feature interaction and fusion, resulting in one unified facial representation. Additionally, we develop one concise face mask mechanism that facilitates multi-view feature fusion and facial reconstruction by identifying common areas and guiding the network’s focus on critical facial features (e.g., eyes, brows, nose, and mouth). Experiments on Pixel-Face and Bosphorus datasets indicate the superiority of the proposed method. Without the 3D annotation, DF-MVR achieves relative 5.2% and 3.0% RMSE improvement over the existing weakly supervised MVRs, respectively, on Pixel-Face and Bosphorus datasets. Our code is available at https://github.com/weiguangzhao/DF_MVR.

3D open-world classification is a challenging yet essential task in dynamic and unstructured real-world scenarios, requiring robust subsequent knowledge adaptation capabilities. While current approaches predominantly rely on 2D pre-trained models through 3D-to-2D projection, their performance degrades severely under arbitrary object orientations. Unlike these present efforts, this work makes a pioneering exploration of 3D generative models for 3D open-world classification-specifically, leverageing the accumulated prior knowledge from these models to provide anchors for novel categories, while integrating a rotation-invariant feature extractor. This innovative synergy endows our pipeline with the advantages of being training-free and pose-invariant, thus well suited to adapt novel categories in 3D open-world classification. Extensive experiments on benchmark datasets demonstrate the potential of this pipeline, achieving state-of-the-art performance on ModelNet10‡ and McGill‡ with 32.7% and 8.7% overall accuracy improvement, respectively. The code is available in the supplementary materials.

This work presents KDTalker++, a real-time system for generating talking portrait videos from a single image using audio or text input. Built on a keypoint-based spatiotemporal diffusion model, it adds voice cloning, background editing, and fine-grained expression control. The demo is available at https://kdtalker.com. A live presentation video is available at https://drive.google.com/file/d/1N4Ggu0Y32DTsV3mbKhXS4kGY6l2ZYKSp/view.

Math reasoning has been one crucial ability of large language models (LLMs), where significant advancements have been achieved in recent years. However, most efforts focus on LLMs by curating high-quality annotation data and intricate training (or inference) paradigms, while the math reasoning performance of multi-modal LLMs (MLLMs) remains lagging behind. Since the MLLM typically consists of an LLM and vision block, we wonder: \textit{Can MLLMs directly absorb math reasoning abilities from off-the-shelf math LLMs without tuning?} Recent model-merging approaches may offer insights into this question. However, they overlook the alignment between the MLLM and LLM, where we find that there is a large gap between their parameter spaces, resulting in lower performance. Our empirical evidence reveals two key factors behind this issue: the identification of crucial reasoning-associated layers in the model and the mitigation of the gaps in parameter space. Based on the empirical insights, we propose \textbf{IP-Merging} that first \textbf{I}dentifies the reasoning-associated parameters in both MLLM and Math LLM, then \textbf{P}rojects them into the subspace of MLLM aiming to maintain the alignment, finally merges parameters in this subspace. IP-Merging is a tuning-free approach since parameters are directly adjusted. Extensive experiments demonstrate that our IP-Merging method can enhance the math reasoning ability of MLLMs directly from Math LLMs without compromising their other capabilities.

Document dewarping aims to rectify deformations in photographic document images, thus improving text readability, which has attracted much attention and made great progress, but it is still challenging to preserve document structures. Given recent advances in diffusion models, it is natural for us to consider their potential applicability to document dewarping. However, it is far from straightforward to adopt diffusion models in document dewarping due to their unfaithful control on highly complex document images (e.g., 2000 × 3000 resolution). In this paper, we propose DvD, the first generative model to tackle document Dewarping via a Diffusion framework. To be specific, DvD introduces a coordinate-level denoising instead of typical pixel-level denoising, generating a mapping for deformation rectification. In addition, we further propose a time-variant condition refinement mechanism to enhance the preservation of document structures. In experiments, we find that current document dewarping benchmarks can not evaluate dewarping models comprehensively. To this end, we present AnyPhotoDoc6300, a rigorously designed large-scale document dewarping benchmark comprising 6,300 real image pairs across three distinct domains, enabling fine-grained evaluation of dewarping models. Comprehensive experiments demonstrate that our proposed DvD can achieve state-of-the-art performance with acceptable computational efficiency on multiple metrics across various benchmarks, including DocUNet, DIR300, and AnyPhotoDoc6300. The new benchmark and code will be publicly available at https://github.com/hanquansanren/DvD.

Open compound domain adaptation (OCDA) aims to transfer knowledge from a labeled source domain to a mix of unlabeled homogeneous compound target domains while generalizing to open unseen domains. Existing OCDA methods solve the intradomain gaps by a divide-and-conquer strategy, which decomposes the problem into several individual and parallel domain adaptation (DA) tasks. In this work, starting from the general DA theory, we establish a novel generalization bound for the setting of OCDA. Built upon this, we argue that conventional OCDA approaches may substantially underestimate the inherent variance inside the compound target domains for model generalization, constraining the model's performance. We subsequently present stochastic compound mixing (SCMix), an augmentation strategy with the primary objective of mitigating the divergence between the source and mixed target distributions. Theoretical analyses are conducted to substantiate the superiority of SCMix, proving that single-target mixing is a subgroup of our method. Extensive experiments show that our method attains a lower empirical risk on OCDA semantic segmentation tasks, thus supporting our theories. In particular, combining the transformer architecture, SCMix achieves a notable performance boost compared to SoTA results.

While recent test-time adaptations exhibit efficacy by adjusting batch normalization to narrow domain disparities, their effectiveness diminishes with realistic mini-batches due to inaccurate target estimation. As previous attempts merely introduce source statistics to mitigate this issue, the fundamental problem of inaccurate target estimation still persists, leaving the intrinsic test-time domain shifts unresolved. This paper delves into the problem of mini-batch degradation. By unraveling batch normalization, we discover that the inexact target statistics largely stem from the substantially reduced class diversity in batch. Drawing upon this insight, we introduce a straightforward tool, Test-time Exponential Moving Average (TEMA), to bridge the class diversity gap between training and testing batches. Importantly, our TEMA adaptively extends the scope of typical methods beyond the current batch to incorporate a diverse set of class information, which in turn boosts an accurate target estimation. Built upon this foundation, we further design a novel layer-wise rectification strategy to consistently promote test-time performance. Our proposed method enjoys a unique advantage as it requires neither training nor tuning parameters, offering a truly hassle-free solution. It significantly enhances model robustness against shifted domains and maintains resilience in diverse real-world scenarios with various batch sizes, achieving state-of-the-art performance on several major benchmarks. Code is available at https://github.com/kiwi12138/RealisticTTA.

Data augmentation has been recently leveraged as an effective regularizer in various vision-language deep neural networks. However, in text-to-image synthesis (T2Isyn), current augmentation wisdom still suffers from the semantic mismatch between augmented paired data. Even worse, semantic collapse may occur when generated images are less semantically constrained. In this paper, we develop a novel Semantic-aware Data Augmentation (SADA) framework dedicated to T2Isyn. In particular, we propose to augment texts in the semantic space via an Implicit Textual Semantic Preserving Augmentation, in conjunction with a specifically designed Image Semantic Regularization Loss as Generated Image Semantic Conservation, to cope well with semantic mismatch and collapse. As one major contribution, we theoretically show that Implicit Textual Semantic Preserving Augmentation can certify better text-image consistency while Image Semantic Regularization Loss regularizing the semantics of generated images would avoid semantic collapse and enhance image quality. Extensive experiments validate that SADA enhances text-image consistency and improves image quality significantly in T2Isyn models across various backbones. Especially, incorporating SADA during the tuning process of Stable Diffusion models also yields performance improvements.

With the boom of Large Language Models (LLMs), the research of solving Math Word Problem (MWP) has recently made great progress. However, there are few studies to examine the robustness of LLMs in math solving ability. Instead of attacking prompts in the use of LLMs, we propose a MathAttack model to attack MWP samples which are closer to the essence of robustness in solving math problems. Compared to traditional text adversarial attack, it is essential to preserve the mathematical logic of original MWPs during the attacking. To this end, we propose logical entity recognition to identify logical entries which are then frozen. Subsequently, the remaining text are attacked by adopting a word-level attacker. Furthermore, we propose a new dataset RobustMath to evaluate the robustness of LLMs in math solving ability. Extensive experiments on our RobustMath and two another math benchmark datasets GSM8K and MultiAirth show that MathAttack could effectively attack the math solving ability of LLMs. In the experiments, we observe that (1) Our adversarial samples from higher-accuracy LLMs are also effective for attacking LLMs with lower accuracy (e.g., transfer from larger to smaller-size LLMs, or from few-shot to zero-shot prompts); (2) Complex MWPs (such as more solving steps, longer text, more numbers) are more vulnerable to attack; (3) We can improve the robustness of LLMs by using our adversarial samples in few-shot prompts. Finally, we hope our practice and observation can serve as an important attempt towards enhancing the robustness of LLMs in math solving ability. The code and dataset is available at: https://github.com/zhouzihao501/MathAttack.

Brain tumor segmentation is often based on multiple magnetic resonance imaging (MRI). However, in clinical practice, certain modalities of MRI may be missing, which presents an even more difficult scenario. To cope with this challenge, knowledge distillation has emerged as one promising strategy. However, recent efforts typically overlook the modality gaps and thus fail to learn invariant feature representations across different modalities. Such drawback consequently leads to limited performance for both teachers and students. To ameliorate these problems, in this paper, we propose a novel paradigm that aligns latent features of involved modalities to a well-defined distribution anchor. As a major contribution, we prove that our novel training paradigm ensures a tight evidence lower bound, thus theoretically certifying its effectiveness. Extensive experiments on different backbones validate that the proposed paradigm can enable invariant feature representations and produce a teacher with narrowed modality gaps. This further offers superior guidance for missing modality students, achieving an average improvement of 1.75 on dice score.

Multi-domain generalization$(mDG)$is universally aimed to minimize the discrepancy between training and testing distributions to enhance marginal-to-label distribution mapping. However, existing$mDG$literature lacks a general learning objective paradigm and often imposes constraints on static target marginal distributions. In this paper, we propose to leverage a Y-mapping to relax the constraint. We rethink the learning objective for$mDG$and design a new general learning objective to interpret and analyze most existing$mDG$wisdom. This general objective is bifurcated into two synergistic amis: learning domain-independent conditional features and maximizing a posterior. Explorations also extend to two effective regularization terms that incorporate prior information and suppress invalid causality, alleviating the issues that come with relaxed constraints. We theoretically contribute an upper bound for the domain alignment of domain-independent conditional features, disclosing that many previous$mDG$endeavors actually optimize partially the objective and thus lead to limited performance. As such, our study distills a general learning objective into four practical components, providing a general, robust, and flexible mechanism to handle complex domain shifts. Extensive empirical results indicate that the proposed objective with Y -mapping leads to substantially better$mDG$performance in various downstream tasks, including regression, segmentation, and classification. Code is available at htttps://github.com/zhaorui-t.an/GMDG/tree/main.

Not relying on ground-truth data for training, self-supervised semantic visual odometry (SVO) has recently gained considerable attention. Within self-supervised SVO, feature representation inconsistency between semantic/depth and pose tasks presents a significant challenge, as it may disrupt cross-task feature representations and lead to notable performance degradation. Regrettably, existing self-supervised SVO lacks an effective solution to address this obstacle, for either overlooking this issue or exploiting a too heavy architecture. In response to this challenge, we propose a groundbreaking solution within the Single-Stream architecture, known as Lite-SVO, which is a lightweight yet efficient multi-feature sharing architecture. Lite-SVO is designed to bolster self-supervised SVO, facilitating its adoption on edge devices without compromising accuracy and performance. The crucial innovation lies in the multi-feature sharing architecture, which fuses the semantic and depth maps as pose features, thus significantly reducing the model complexity and boosting the speed in edge devices. Built upon the novel feature sharing framework, Lite-SVO further optimizes the feature sharing representation to improve the performance. Specifically, a cross-feature sharing module alleviates the impact of object boundary in depth estimation, while a multi-feature sharing module focuses on extracting and fusing spatial features to enhance pose estimation. Experimental results demonstrate that our method is at least 84.46% faster than the state-of-the-art Single-Stream approaches, and excitingly, our method’s pose accuracy is about 79.83% higher than theirs.

Vision models excel in image classification but struggle to generalize to unseen data, such as classifying images from unseen domains or discovering novel categories. In this paper, we explore the relationship between logical reasoning and deep learning generalization in visual classification. A logical regularization termed L-Reg is derived which bridges a logical analysis framework to image classification. Our work reveals that L-Reg reduces the complexity of the model in terms of the feature distribution and classifier weights. Specifically, we unveil the interpretability brought by L-Reg, as it enables the model to extract the salient features, such as faces to persons, for classification. Theoretical analysis and experiments demonstrate that L-Reg enhances generalization across various scenarios, including multi-domain generalization and generalized category discovery. In complex real-world scenarios where images span unknown classes and unseen domains, L-Reg consistently improves generalization, highlighting its practical efficacy.

Zero-shot learning (ZSL) refers to the design of predictive functions on new classes (unseen classes) of data that have never been seen during training. In a more practical scenario, generalized zero-shot learning (GZSL) requires predicting both seen and unseen classes accurately. In the absence of target samples, many GZSL models may overfit training data and are inclined to predict individuals as categories that have been seen in training. To alleviate this problem, we develop a parameter-wise adversarial training process that promotes robust recognition of seen classes while designing during the test a novel model perturbation mechanism to ensure sufficient sensitivity to unseen classes. Concretely, adversarial perturbation is conducted on the model to obtain instance-specific parameters so that predictions can be biased to unseen classes in the test. Meanwhile, the robust training encourages the model robustness, leading to nearly unaffected prediction for seen classes. Moreover, perturbations in the parameter space, computed from multiple individuals simultaneously, can be used to avoid the effect of perturbations that are too extreme and ruin the predictions. Comparison results on four benchmark ZSL data sets show the effective improvement that the proposed framework made on zero-shot methods with learned metrics.

While Graph Neural Networks (GNNs) have achieved enormous success in multiple graph analytical tasks, modern variants mostly rely on the strong inductive bias of homophily. However, real-world networks typically exhibit both homophilic and heterophilic linking patterns, wherein adjacent nodes may share dissimilar attributes and distinct labels. Therefore, GNNs smoothing node proximity holistically may aggregate both task-relevant and irrelevant (even harmful) information, limiting their ability to generalize to heterophilic graphs and potentially causing non-robustness. In this work, we propose a novel Edge Splitting GNN (ES-GNN) framework to adaptively distinguish between graph edges either relevant or irrelevant to learning tasks. This essentially transfers the original graph into two subgraphs with the same node set but complementary edge sets dynamically. Given that, information propagation separately on these subgraphs and edge splitting are alternatively conducted, thus disentangling the task-relevant and irrelevant features. Theoretically, we show that our ES-GNN can be regarded as a solution to a disentangled graph denoising problem, which further illustrates our motivations and interprets the improved generalization beyond homophily. Extensive experiments over 11 benchmark and 1 synthetic datasets not only demonstrate the effective performance of ES-GNN but also highlight its robustness to adversarial graphs and mitigation of the over-smoothing problem.

While exogenous variables have a major impact on performance improvement in time series analysis, interseries correlation and time dependence among them are rarely considered in the present continuous methods. The dynamical systems of multivariate time series could be modeled with complex unknown partial differential equations (PDEs) which play a prominent role in many disciplines of science and engineering. In this article, we propose a continuous-time model for arbitrary-step prediction to learn an unknown PDE system in multivariate time series whose governing equations are parameterized by self-attention and gated recurrent neural networks. The proposed model, exogenous-guided PDE network (EgPDE-Net), takes account of the relationships among the exogenous variables and their effects on the target series. Importantly, the model can be reduced into a regularized ordinary differential equation (ODE) problem with specially designed regularization guidance, which makes the PDE problem tractable to obtain numerical solutions and feasible to predict multiple future values of the target series at arbitrary time points. Extensive experiments demonstrate that our proposed model could achieve competitive accuracy over strong baselines: on average, it outperforms the best baseline by reducing 9.85% on RMSE and 13.98% on MAE for arbitrary-step prediction.

Graph convolutional networks (GCNs) emerge as the most successful learning models for graph-structured data. Despite their success, existing GCNs usually ignore the entangled latent factors typically arising in real-world graphs, which results in nonexplainable node representations. Even worse, while the emphasis has been placed on local graph information, the global knowledge of the entire graph is lost to a certain extent. In this work, to address these issues, we propose a novel framework for GCNs, termed LGD-GCN, taking advantage of both local and global information for disentangling node representations in the latent space. Specifically, we propose to represent a disentangled latent continuous space with a statistical mixture model, by leveraging neighborhood routing mechanism locally. From the latent space, various new graphs can then be disentangled and learned, to overall reflect the hidden structures with respect to different factors. On the one hand, a novel regularizer is designed to encourage interfactor diversity for model expressivity in the latent space. On the other hand, the factor-specific information is encoded globally via employing a message passing along these new graphs, in order to strengthen intrafactor consistency. Extensive evaluations on both synthetic and five benchmark datasets show that LGD-GCN brings significant performance gains over the recent competitive models in both disentangling and node classification. Particularly, LGD-GCN is able to outperform averagely the disentangled state-of-the-arts by 7.4% on social network datasets.

In the stock market, a successful investment requires a good balance between profits and risks. Based on the learning to rank paradigm, stock recommendation has been widely studied in quantitative finance to recommend stocks with higher return ratios for investors. Despite the efforts to make profits, many existing recommendation approaches still have some limitations in risk control, which may lead to intolerable paper losses in practical stock investing. To effectively reduce risks, we draw inspiration from adversarial learning and propose a novel Split Variational Adversarial Training (SVAT) method for risk-aware stock recommendation. Essentially, SVAT encourages the stock model to be sensitive to adversarial perturbations of risky stock examples and enhances the model’s risk awareness by learning from perturbations. To generate representative adversarial examples as risk indicators, we devise a variational perturbation generator to model diverse risk factors. Particularly, the variational architecture enables our method to provide a rough risk quantification for investors, showing an additional advantage of interpretability. Experiments on several real-world stock market datasets demonstrate the superiority of our SVAT method. By lowering the volatility of the stock-recommendation model, SVAT effectively reduces investment risks and outperforms state-of-the-art baselines by more than 30% in terms of risk-adjusted profits. All the experimental data and source code are available at https://drive.google.com/drive/folders/14AdM7WENEvIp5x5bV3zV_i4Aev21C9g6?usp=sharing .

Generalised image outpainting is an important and active research topic in computer vision, which aims to extend appealing content all-side around a given image. Existing state-of-the-art outpainting methods often rely on discrete extrapolation to extend the feature map in the bottleneck. They thus suffer from content unsmoothness, especially in circumstances where the outlines of objects in the extrapolated regions are incoherent with the input sub-images. To mitigate this issue, we design a novel bottleneck with Neural ODEs to make continuous extrapolation in latent space, which could be a plug-in for many deep learning frameworks. Our ODE-based network continuously transforms the state and makes accurate predictions by learning the incremental relationship among latent points, leading to both smooth and structured feature representation. Experimental results on three real-world datasets both applied on transformer-based and CNN-based frameworks show that our methods could generate more realistic and coherent images against the state-of-the-art image outpainting approaches. Our code is available at https://github.com/PengleiGao/Continuous-Image-Outpainting-with-Neural-ODE .

Scene text recognition (STR), one typical sequence-to-sequence problem, has drawn much attention recently in multimedia applications. To guarantee good performance, it is essential for STR to obtain aligned character-wise features from the whole-image feature maps. While most present works adopt fully data-driven attention-based alignment, such practice ignores specific character geometric information. In this article, built upon a group of learnable geometric points, we propose a novel shape-driven attention alignment method that is able to obtain character-wise features. Concretely, we first design a corner detector to generate a shape map to guide the attention alignments explicitly, where a series of points can be learned to represent character-wise features flexibly. We then propose a dual-path network with a mutual learning and cooperating strategy that successfully combines CNN with a ViT-based model, leading to further accuracy improvement. We conduct extensive experiments to evaluate the proposed method on various scene text benchmarks, including six popular regular and irregular datasets, two more challenging datasets (i.e., WordArt and OST), and three Chinese datasets. Experimental results indicate that our method can achieve superior performance with a comparable model size against many state-of-the-art models.

Single-source domain generalization (SDG) in medical image segmentation is a challenging yet essential task as domain shifts are quite common among clinical image datasets. Previous attempts most conduct global-only/random augmentation. Their augmented samples are usually insufficient in diversity and informativeness, thus failing to cover the possible target domain distribution. In this paper, we rethink the data augmentation strategy for SDG in medical image segmentation. Motivated by the class-level representation invariance and style mutability of medical images, we hypothesize that unseen target data can be sampled from a linear combination of C (the class number) random variables, where each variable follows a location-scale distribution at the class level. Accordingly, data augmented can be readily made by sampling the random variables through a general form. On the empirical front, we implement such strategy with constrained Bezier transformation on both global and local (i.e. class-level) regions, which can largely increase the augmentation diversity. A Saliency-balancing Fusion mechanism is further proposed to enrich the informativeness by engaging the gradient information, guiding augmentation with proper orientation and magnitude. As an important contribution, we prove theoretically that our proposed augmentation can lead to an upper bound of the generalization risk on the unseen target domain, thus confirming our hypothesis. Combining the two strategies, our Saliency-balancing Location-scale Augmentation (SLAug) exceeds the state-of-the-art works by a large margin in two challenging SDG tasks. Code is available at https://github.com/Kaiseem/SLAug.

In domain adaptive segmentation, domain shift may cause erroneous high-confidence predictions on the target domain, resulting in poor self-training. To alleviate the potential error, most previous works mainly consider aleatoric uncertainty arising from the inherit data noise. This may however lead to overconfidence in incorrect predictions and thus limit the performance. In this paper, we take advantage of Deterministic Uncertainty Methods (DUM) to explore the epistemic uncertainty, which reflects accurately the domain gap depending on the model choice and parameter fitting trained on source domain. The epistemic uncertainty on target domain is evaluated on-the-fly to facilitate online reweighting and correction in the self-training process. Meanwhile, to tackle the class-wise quantity and learning difficulty imbalance problem, we introduce a novel data resampling strategy to promote simultaneous convergence across different categories. This strategy prevents the class-level over-fitting in source domain and further boosts the adaptation performance by better quantifying the uncertainty in target domain. We illustrate the superiority of our method compared with the state-of-the-art methods.

Recent studies have investigated how to achieve robustness for unsupervised domain adaptation (UDA). While most efforts focus on adversarial robustness, i.e. how the model performs against unseen malicious adversarial perturbations, robustness against benign common corruption (RaCC) surprisingly remains under-explored for UDA. Towards improving RaCC for UDA methods in an unsupervised manner, we propose a novel Distributionally and Discretely Adversarial Regularization (DDAR) framework in this paper. Formulated as a min-max optimization with a distribution distance, DDAR1is theoretically well-founded to ensure generalization over unknown common corruptions. Meanwhile, we show that our regularization scheme effectively reduces a surrogate of RaCC, i.e., the perceptual distance between natural data and common corruption. To enable a abetter adversarial regularization, the design of the optimization pipeline relies on an image discretization scheme that can transform "out-of-distribution" adversarial data into "in-distribution" data augmentation. Through extensive experiments, in terms of RaCC, our method is superior to conventional unsupervised regularization mechanisms, widely improves the robustness of existing UDA methods, and achieves state-of-the-art performance.

Instance segmentation on point clouds is crucially important for 3D scene understanding. Most SOTAs adopt distance clustering, which is typically effective but does not perform well in segmenting adjacent objects with the same semantic label (especially when they share neighboring points). Due to the uneven distribution of offset points, these existing methods can hardly cluster all instance points. To this end, we design a novel divide-and-conquer strategy named PBNet that binarizes each point and clusters them separately to segment instances. Our binary clustering divides offset instance points into two categories: high and low density points (HPs vs. LPs). Adjacent objects can be clearly separated by removing LPs, and then be completed and refined by assigning LPs via a neighbor voting method. To suppress potential over-segmentation, we propose to construct local scenes with the weight mask for each instance. As a plug-in, the proposed binary clustering can replace the traditional distance clustering and lead to consistent performance gains on many mainstream baselines. A series of experiments on ScanNetV2 and S3DIS datasets indicate the superiority of our model. In particular, PBNet ranks first on the ScanNetV2 official benchmark challenge, achieving the highest mAP. Code will be available publicly at https://github.com/weiguangzhao/PBNet.

Panoramic Narrative Grounding (PNG) is an emerging visual grounding task that aims to segment visual objects in images based on dense narrative captions. The current state-of-the-art methods first refine the representation of phrase by aggregating the most similar k image pixels, and then match the refined text representations with the pixels of the image feature map to generate segmentation results. However, simply aggregating sampled image features ignores the contextual information, which can lead to phrase-to-pixel mis-match. In this paper, we propose a novel learning framework called Deformable Attention Refined Matching Network (DRMN), whose main idea is to bring deformable attention in the iterative process of feature learning to incorporate essential context information of different scales of pixels. DRMN iteratively re-encodes pixels with the deformable attention network after updating the feature representation of the top-k most similar pixels. As such, DRMN can lead to accurate yet discriminative pixel representations, purify the top-k most similar pixels, and consequently alleviate the phrase-to-pixel mis-match substantially. Experimental results show that our novel design significantly improves the matching results between text phrases and image pixels. Concretely, DRMN achieves new state-of-the-art performance on the PNG benchmark with an average recall improvement 3.5%. The codes are available in: https://github.com/JaMesLiMers/DRMN.

Recent deep generative models have achieved promising performance in image inpainting. However, it is still challenging for a neural network to generate realistic image details and textures due to its inherent spectral bias. We suggest adopting a ‘structure first detail next’ workflow for image inpainting by knowing how artists work. Thus, we propose to build a Pyramid Generator by stacking several sub-generators, where lower-layer sub-generators focus on restoring image structures. In contrast, the higher-layer sub-generators emphasize image details. Our model progressively restores the input through the entire pyramid in a bottom-up fashion. Notably, our approach has a learning scheme of progressively increasing hole size, which allows it to restore large-hole images. In addition, our method could fully exploit the benefits of learning with high-resolution images and hence is suitable for high-resolution image inpainting. Extensive experimental results on benchmark datasets have validated the effectiveness of our approach compared with state-of-the-arts.

AI has made significant progress in solving math problems, but geometry problems remain challenging due to their reliance on both text and diagrams. In the text description, symbolic characters such as "ABC" often serve as a bridge to connect the corresponding diagram. However, by simply tokenizing symbolic characters into individual letters (e.g., 'A', 'B' and 'C'), existing works fail to study them explicitly and thus lose the semantic relationship with the diagram. In this paper, we develop a symbolic character-aware model to fully explore the role of these characters in both text and diagram understanding and optimize the model under a multi-modal reasoning framework. In the text encoder, we propose merging individual symbolic characters to form one semantic unit along with geometric information from the corresponding diagram. For the diagram encoder, we pre-train it under a multi-label classification framework with the symbolic characters as labels. In addition, we enhance the geometry diagram understanding ability via a self-supervised learning method under the masked image modeling auxiliary task. By integrating the proposed model into a general encoder-decoder pipeline for solving geometry problems, we demonstrate its superiority on two benchmark datasets, including GeoQA and Geometry3K, with extensive experiments. Specifically, on GeoQA, the question-solving accuracy is increased from 60.0% to 64.1%, achieving a new state-of-the-art accuracy; on Geometry3K, we reduce the question average solving steps from 6.9 down to 6.0 with marginally higher solving accuracy.

Spectral Graph Neural Networks (GNNs) have achieved tremendous success in graph machine learning, with polynomial filters applied for graph convolutions, where all nodes share the identical filter weights to mine their local contexts. Despite the success, existing spectral GNNs usually fail to deal with complex networks (e.g., WWW) due to such homogeneous spectral filtering setting that ignores the regional heterogeneity as typically seen in real-world networks. To tackle this issue, we propose a novel diverse spectral filtering (DSF) framework, which automatically learns node-specific filter weights to exploit the varying local structure properly. Particularly, the diverse filter weights consist of two components — A global one shared among all nodes, and a local one that varies along network edges to reflect node difference arising from distinct graph parts — to balance between local and global information. As such, not only can the global graph characteristics be captured, but also the diverse local patterns can be mined with awareness of different node positions. Interestingly, we formulate a novel optimization problem to assist in learning diverse filters, which also enables us to enhance any spectral GNNs with our DSF framework. We showcase the proposed framework on three state-of-the-arts including GPR-GNN, BernNet, and JacobiConv. Extensive experiments over 10 benchmark datasets demonstrate that our framework can consistently boost model performance by up to 4.92% in node classification tasks, producing diverse filters with enhanced interpretability.

Deciphering oracle bone script is of great significance to the study of ancient Chinese culture as well as archaeology. Although recent studies on oracle character recognition have made substantial progress, they still suffer from the long-tailed data situation that results in a noticeable performance drop on the tail classes. To mitigate this issue, we propose a generative adversarial framework to augment oracle characters in the problematic classes. In this framework, the generator produces synthetic data through convex combinations of all the available samples in the corresponding classes, and is further optimized through adversarial learning with the classifier and simultaneously the discriminator . Meanwhile, we introduce Repatch to generalize samples in the generator. Since tail classes do not have sufficient data for convex combinations , we propose the TailMix mechanism to generate suitable tail class samples from other classes. Experimental results show that our proposed algorithm obtains remarkable performance in oracle character recognition and achieves new state-of-the-art average (total) accuracy with 86.03% (89.46%), 86.54% (93.86%), 95.22% (96.17%) on the three datasets Oracle-AYNU, OBC306 and Oracle-20K, respectively.

Zero-shot learning (ZSL) aims to identify unseen classes with zero samples during training. Broadly speaking, present ZSL methods usually adopt class-level semantic labels and compare them with instance-level semantic predictions to infer unseen classes. However, we find that such existing models mostly produce imbalanced semantic predictions, i.e. these models could perform precisely for some semantics, but may not for others. To address the drawback, we aim to introduce an imbalanced learning framework into ZSL. However, we find that imbalanced ZSL has two unique challenges: (1) Its imbalanced predictions are highly correlated with the value of semantic labels rather than the number of samples as typically considered in the traditional imbalanced learning; (2) Different semantics follow quite different error distributions between classes. To mitigate these issues, we first formalize ZSL as an imbalanced regression problem which offers empirical evidences to interpret how semantic labels lead to imbalanced semantic predictions. We then propose a re-weighted loss termed Re-balanced Mean-Squared Error (ReMSE), which tracks the mean and variance of error distributions, thus ensuring rebalanced learning across classes. As a major contribution, we conduct a series of analyses showing that ReMSE is theoretically well established. Extensive experiments demonstrate that the proposed method effectively alleviates the imbalance in semantic prediction and outperforms many state-of-the-art ZSL methods.

Unsupervised cross-modality medical image adaptation aims to alleviate the severe domain gap between different imaging modalities without using the target domain label. A key in this campaign relies upon aligning the distributions of source and target domain. One common attempt is to enforce the global alignment between two domains, which, however, ignores the fatal local-imbalance domain gap problem, i.e., some local features with larger domain gap are harder to transfer. Recently, some methods conduct alignment focusing on local regions to improve the efficiency of model learning. While this operation may cause a deficiency of critical information from contexts. To tackle this limitation, we propose a novel strategy to alleviate the domain gap imbalance considering the characteristics of medical images, namely Global-Local Union Alignment. Specifically, a feature-disentanglement style-transfer module first synthesizes the target-like source images to reduce the global domain gap. Then, a local feature mask is integrated to reduce the 'inter-gap' for local features by prioritizing those discriminative features with larger domain gap. This combination of global and local alignment can precisely localize the crucial regions in segmentation target while preserving the overall semantic consistency. We conduct a series of experiments with two cross-modality adaptation tasks, i,e. cardiac substructure and abdominal multi-organ segmentation. Experimental results indicate that our method achieves state-of-the-art performance in both tasks.

We focus on multi-step ahead time series forecasting with the multi-output strategy. From the perspective of multi-task learning (MTL), we recognize imbalanced uncertainties between prediction tasks of different future time steps. Unexpectedly, trained by the standard summed Mean Squared Error (MSE) loss, existing multi-output forecasting models may suffer from performance drops due to the inconsistency between the loss function and the imbalance structure. To address this problem, we reformulate each prediction task as a distinct Gaussian Mixture Model (GMM) and derive a multi-level Gaussian mixture loss function to better fit imbalanced uncertainties in multi-output time series forecasting. Instead of using the two-step Expectation-Maximization (EM) algorithm, we apply the self-attention mechanism on the task-specific parameters to learn the correlations between different prediction tasks and generate the weight distribution for each GMM component. In this way, our method jointly optimizes the parameters of the forecasting model and the mixture model simultaneously in an end-to-end fashion, avoiding the need of two-step optimization. Experiments on three real-world datasets demonstrate the effectiveness of our multi-level Gaussian mixture loss compared to models trained with the standard summed MSE loss function. All the experimental data and source code are available at https://github.com/smallGum/GMM-FNN .

In this work, we propose a continuous neural network architecture, referred to as Explainable Tensorized Neural - Ordinary Differential Equations (ETN-ODE) network for multi-step time series prediction at arbitrary time points. Unlike existing approaches which mainly handle univariate time series for multi-step prediction, or multivariate time series for single-step predictions, ETN-ODE is capable of handling multivariate time series with arbitrary-step predictions. An additional benefit is its tandem attention mechanism, with respect to temporal and variable attention, which enable it to greatly facilitate data interpretability. Specifically, the proposed model combines an explainable tensorized gated recurrent unit with ordinary differential equations, with the derivatives of the latent states parameterized through a neural network. We quantitatively and qualitatively demonstrate the effectiveness and interpretability of ETN-ODE on one arbitrary-step prediction task and five standard multi-step prediction tasks. Extensive experiments show that the proposed method achieves very accurate predictions at arbitrary time points while attaining very competitive performance against the baseline methods in standard multi-step time series prediction.

AdaBelief, one of the current best optimizers, demonstrates superior generalization ability over the popular Adam algorithm by viewing the exponential moving average of observed gradients. AdaBelief is theoretically appealing in which it has a data-dependent O(√T) regret bound when objective functions are convex, where T is a time horizon. It remains, however, an open problem whether the convergence rate can be further improved without sacrificing its generalization ability. To this end, we make the first attempt in this work and design a novel optimization algorithm called FastAdaBelief that aims to exploit its strong convexity in order to achieve an even faster convergence rate. In particular, by adjusting the step size that better considers strong convexity and prevents fluctuation, our proposed FastAdaBelief demonstrates excellent generalization ability and superior convergence. As an important theoretical contribution, we prove that FastAdaBelief attains a data-dependent O(logT) regret bound, which is substantially lower than AdaBelief in strongly convex cases. On the empirical side, we validate our theoretical analysis with extensive experiments in scenarios of strong convexity and nonconvexity using three popular baseline models. Experimental results are very encouraging: FastAdaBelief converges the quickest in comparison to all mainstream algorithms while maintaining an excellent generalization ability, in cases of both strong convexity or nonconvexity. FastAdaBelief is, thus, posited as a new benchmark model for the research community.

Recent studies explore how to obtain adversarial robustness for unsupervised domain adaptation (UDA). These efforts are however dedicated to achieving an optimal trade-off between accuracy and robustness on a given or seen target domain but ignore the robust generalization issue over unseen adversarial data. Consequently, degraded performance will be often observed when existing robust UDAs are applied to future adversarial data. In this work, we make a first attempt to address the robust generalization issue of UDA. We conjecture that the poor robust generalization of present robust UDAs may be caused by the large distribution gap among adversarial examples. We then provide an empirical and theoretical analysis showing that this large distribution gap is mainly owing to the discrepancy between feature-shift distributions. To reduce such discrepancy, a novel Anchored Feature-Shift Regularization (AFSR) method is designed with a certificated robust generalization bound. We conduct a series of experiments on benchmark UDA datasets. Experimental results validate the effectiveness of our proposed AFSR over many existing robust UDA methods.

Text classification techniques have been substantially important to many smart computing applications, e.g. topic extraction and event detection. However, classification is always challenging when only insufficient amount of labelled data for model training is available. To mitigate this issue, zero-shot learning (ZSL) has been introduced for models to recognise new classes that have not been observed during the training stage. We propose an entailment-based zero-shot text classification model, named as S-BERT-CAM, to better capture the relationship between the premise and hypothesis in the BERT embedding space. Two widely used textual datasets are utilised to conduct the experiments. We fine-tune our model using 50% of the labels for each dataset and evaluate it on the label space containing all labels (including both seen and unseen labels). The experimental results demonstrate that our model is more robust to the generalised ZSL and significantly improves the overall performance against baselines.

The idea of “citizen sensing” and “human as sensors” is crucial for social Internet of Things, an integral part of cyber–physical–social systems (CPSSs). Social media data, which can be easily collected from the social world, has become a valuable resource for research in many different disciplines, e.g., crisis/disaster assessment, social event detection, or the recent COVID-19 analysis. Useful information, or knowledge derived from social data, could better serve the public if it could be processed and analyzed in more efficient and reliable ways. Advances in deep neural networks have significantly improved the performance of many social media analysis tasks. However, deep learning models typically require a large amount of labeled data for model training, while most CPSS data is not labeled, making it impractical to build effective learning models using traditional approaches. In addition, the current state-of-the-art, pretrained natural language processing (NLP) models do not make use of existing knowledge graphs, thus often leading to unsatisfactory performance in real-world applications. To address the issues, we propose a new zero-shot learning method which makes effective use of existing knowledge graphs for the classification of very large amounts of social text data. Experiments were performed on a large, real-world tweet data set related to COVID-19, the evaluation results show that the proposed method significantly outperforms six baseline models implemented with state-of-the-art deep learning models for NLP.

We consider the problem of volumetric (3D) unsupervised domain adaptation (UDA) in cross-modality medical image segmentation, aiming to perform segmentation on the unannotated target domain (e.g. MRI) with the help of labeled source domain (e.g. CT). Previous UDA methods in medical image analysis usually suffer from two challenges: 1) they focus on processing and analyzing data at 2D level only, thus missing semantic information from the depth level; 2) one-to-one mapping is adopted during the style-transfer process, leading to insufficient alignment in the target domain. Different from the existing methods, in our work, we conduct a first of its kind investigation on multi-style image translation for complete image alignment to alleviate the domain shift problem, and also introduce 3D segmentation in domain adaptation tasks to maintain semantic consistency at the depth level. In particular, we develop an unsupervised domain adaptation framework incorporating a novel quartet self-attention module to efficiently enhance relationships between widely separated features in spatial regions on a higher dimension, leading to a substantial improvement in segmentation accuracy in the unlabeled target domain. In two challenging cross-modality tasks, specifically brain structures and multi-organ abdominal segmentation, our model is shown to outperform current state-of-the-art methods by a significant margin, demonstrating its potential as a benchmark resource for the biomedical and health informatics research community.

Using generative models to synthesize visual features from semantic distribution is one of the most popular solutions to ZSL image classification in recent years. The triplet loss (TL) is popularly used to generate realistic visual distributions from semantics by automatically searching discriminative representations. However, the traditional TL cannot search reliable unseen disentangled representations due to the unavailability of unseen classes in ZSL. To alleviate this drawback, we propose in this work a multi-modal triplet loss (MMTL) which utilizes multi-modal information to search adisentangledrepresentation space. As such, all classes can interplay which can benefit learning disentangled class representations in the searched space. Furthermore, we develop a novel model called Disentangling Class Representation Generative Adversarial Network (DCR-GAN) focusing on exploiting the disentangled representations in training, feature synthesis, and final recognition stages. Benefiting from the disentangled representations, DCR-GAN could fit a more realistic distribution over both seen and unseen features. Extensive experiments show that our proposed model can lead to superior performance to the state-of-the-arts on four benchmark datasets.

Currently, many action recognition methods mostly consider the information from spatial streams. We propose a new perspective inspired by the human visual system to combine both spatial and temporal streams to measure their attention consistency. Specifically, a branch-independent convolutional neural network (CNN) based algorithm is developed with a novel attention-consistency loss metric, enabling the temporal stream to concentrate on consistent discriminative regions with the spatial stream in the same period. The consistency loss is further combined with the cross-entropy loss to enhance the visual attention consistency. We evaluate the proposed method for action recognition on two benchmark datasets: Kinetics400 and UCF101. Despite its apparent simplicity, our proposed framework with the attention consistency achieves better performance than most of the two-stream networks, i.e., 75.7% top-1 accuracy on Kinetics400 and 95.7% on UCF101, while reducing 7.1% computational cost compared with our baseline. Particularly, our proposed method can attain remarkable improvements on complex action classes, showing that our proposed network can act as a potential benchmark to handle complicated scenarios in industry 4.0 applications.

The latest heuristic for handling the domain shift in un-supervised domain adaptation tasks is to reduce the data distribution discrepancy using adversarial learning. Recent studies improve the conventional adversarial domain adaptation methods with discriminative information by integrating the classifier’s outputs into distribution divergence measurement. However, they still suffer from the equilibrium problem of adversarial learning in which even if the discriminator is fully confused, sufficient similarity between two distributions cannot be guaranteed. To overcome this problem, we propose a novel approach named feature gradient distribution alignment (FGDA)1. We demonstrate the rationale of our method both theoretically and empirically. In particular, we show that the distribution discrepancy can be reduced by constraining feature gradients of two domains to have similar distributions. Meanwhile, our method enjoys a theoretical guarantee that a tighter error upper bound for target samples can be obtained than that of conventional adversarial domain adaptation methods. By integrating the proposed method with existing adversarial domain adaptation models, we achieve state-of-the-art performance on two real-world benchmark datasets.

While adversarial training becomes one of the most promising defending approaches against adversarial attacks for deep neural networks, the conventional wisdom through robust optimization may usually not guarantee good generalization for robustness. Concerning with robust generalization over unseen adversarial data, this paper investigates adversarial training from a novel perspective of shift consistency in latent space. We argue that the poor robust generalization of adversarial training is owing to the significantly dispersed latent representations generated by training and test adversarial data, as the adversarial perturbations push the latent features of natural examples in the same class towards diverse directions. This is underpinned by the theoretical analysis of the robust generalization gap, which is upper-bounded by the standard one over the natural data and a term of feature inconsistent shift caused by adversarial perturbation {–} a measure of latent dispersion. Towards better robust generalization, we propose a new regularization method {–} shift consistency regularization (SCR) {–} to steer the same-class latent features of both natural and adversarial data into a common direction during adversarial training. The effectiveness of SCR in adversarial training is evaluated through extensive experiments over different datasets, such as CIFAR-10, CIFAR-100, and SVHN, against several competitive methods.

This study develops a calibrated beam-based algorithm with awareness of the global attention distribution for neural abstractive summarization, aiming to improve the local optimality problem of the original beam search in a rigorous way. Specifically, a novel global protocol is proposed based on the attention distribution to stipulate how a global optimal hypothesis should attend to the source. A global scoring mechanism is then developed to regulate beam search to generate summaries in a near-global optimal fashion. This novel design enjoys a distinctive property, i.e., the global attention distribution could be predicted before inference, enabling step-wise improvements on the beam search through the global scoring mechanism. Extensive experiments on nine datasets show that the global (attention)-aware inference significantly improves state-of-the-art summarization models even using empirical hyper-parameters. The algorithm is also proven robust as it remains to generate meaningful texts with corrupted attention distributions. The codes and a comprehensive set of examples are available.

Automated social text annotation is the task of suggesting a set of tags for shared documents on social media platforms. The automated annotation process can reduce users' cognitive overhead in tagging and improve tag management for better search, browsing, and recommendation of documents. It can be formulated as a multilabel classification problem. We propose a novel deep learning-based method for this problem and design an attention-based neural network with semantic-based regularization, which can mimic users' reading and annotation behavior to formulate better document representation, leveraging the semantic relations among labels. The network separately models the title and the content of each document and injects an explicit, title-guided attention mechanism into each sentence. To exploit the correlation among labels, we propose two semantic-based loss regularizers, i.e., similarity and subsumption, which enforce the output of the network to conform to label semantics. The model with the semantic-based loss regularizers is referred to as the joint multilabel attention network (JMAN). We conducted a comprehensive evaluation study and compared JMAN to the state-of-the-art baseline models, using four large, real-world social media data sets. In terms of F1, JMAN significantly outperformed bidirectional gated recurrent unit (Bi-GRU) relatively by around 12.8%-78.6% and the hierarchical attention network (HAN) by around 3.9%-23.8%. The JMAN model demonstrates advantages in convergence and training speed. Further improvement of performance was observed against latent Dirichlet allocation (LDA) and support vector machine (SVM). When applying the semantic-based loss regularizers, the performance of HAN and Bi-GRU in terms of F1was also boosted. It is also found that dynamic update of the label semantic matrices (JMANd) has the potential to further improve the performance of JMAN but at the cost of substantial memory and warrants further study.

Multivariate time series forecasting is an important yet challenging problem in machine learning. Most existing approaches only forecast the series value of one future moment, ignoring the interactions between predictions of future moments with different temporal distance. Such a deficiency probably prevents the model from getting enough information about the future, thus limiting the forecasting accuracy. To address this problem, we propose Multi-Level Construal Neural Network (MLCNN), a novel multi-task deep learning framework. Inspired by the Construal Level Theory of psychology, this model aims to improve the predictive performance by fusing forecasting information (i.e., future visions) of different future time. We first use the Convolution Neural Network to extract multi-level abstract representations of the raw data for near and distant future predictions. We then model the interplay between multiple predictive tasks and fuse their future visions through a modified Encoder-Decoder architecture. Finally, we combine traditional Autoregression model with the neural network to solve the scale insensitive problem. Experiments on three real-world datasets show that our method achieves statistically significant improvements compared to the most state-of-the-art baseline methods, with average 4.59% reduction on RMSE metric and average 6.87% reduction on MAE metric.

Weakly supervised semantic segmentation is a challenging task as it only takes image-level information as supervision for training but produces pixel-level predictions for testing. To address such a challenging task, most recent state-of-the-art approaches propose to adopt two-step solutions, i.e. 1) learn to generate pseudo pixel-level masks, and 2) engage FCNs to train the semantic segmentation networks with the pseudo masks. However, the two-step solutions usually employ many bells and whistles in producing high-quality pseudo masks, making this kind of methods complicated and inelegant. In this work, we harness the image-level labels to produce reliable pixel-level annotations and design a fully end-to-end network to learn to predict segmentation maps. Concretely, we firstly leverage an image classification branch to generate class activation maps for the annotated categories, which are further pruned into confident yet tiny object/background regions. Such reliable regions are then directly served as ground-truth labels for the parallel segmentation branch, where a newly designed dense energy loss function is adopted for optimization. Despite its apparent simplicity, our one-step solution achieves competitive mIoU scores (val: 62.6, test: 62.9) on Pascal VOC compared with those two-step state-of-the-arts. By extending our one-step method to two-step, we get a new state-of-the-art performance on the Pascal VOC (val: 66.3, test: 66.5).

A photovoltaic (PV) generator exhibits nonlinear current-voltage characteristics and its maximum power point varies with incident atmospheric conditions. Therefore, maximum power point tracking (MPPT) control is required to maximize the output power of the PV generator. In this paper, deep Q-network based reinforcement learning strategy is proposed to optimize MPPT process for the photovoltaic system. The proposed system uses a novel control method which introduces agent to interface with the environment and finally gets the strategy of maximum reward accordingly. Simulations and experiments show the feasibility and effectiveness of the proposed system. Compared with the traditional perturb and observe (P&O) and incremental conductance (InC) methods, this method prominently saves tracking steps.

Deep neural network with multi-scale feature fusion has achieved great success in human pose estimation. However, drawbacks still exist in these methods: 1) they consider multi-scale features equally, which may over-emphasize redundant features; 2) preferring deeper structures, they can learn features with the strong semantic representation, but tend to lose natural discriminative information; 3) to attain good performance, they rely heavily on pretraining, which is time-consuming, or even unavailable practically. To mitigate these problems, we propose a novel comprehensive recalibration model called Pyramid GAting Network (PGA-Net) that is capable of distillating, selecting, and fusing the discriminative and attention-aware features at different scales and different levels (i.e., both semantic and natural levels). Meanwhile, focusing on fusing features both selectively and comprehensively, PGA-Net can demonstrate remarkable stability and encouraging performance even without pre-training, making the model can be trained truly from scratch. We demonstrate the effectiveness of PGA-Net through validating on COCO and MPII benchmarks, attaining new state-of-the-art performance. https://github.com/ssr0512/PGA-Net

Social media platforms such as Twitter are increasingly used to collect data of all kinds. During natural disasters, users may post text and image data on social media platforms to report information about infrastructure damage, injured people, cautions and warnings. Effective processing and analysing tweets in real time can help city organisations gain situational awareness of the affected citizens and take timely operations. With the advances in deep learning techniques, recent studies have significantly improved the performance in classifying crisis-related tweets. However, deep learning models are vulnerable to adversarial examples, which may be imperceptible to the human, but can lead to model's misclassification. To process multi-modal data as well as improve the robustness of deep learning models, we propose a multi-modal adversarial training method for crisis-related tweets classification in this paper. The evaluation results clearly demonstrate the advantages of the proposed model in improving the robustness of tweet classification.

Correlation filter has been proven to be an effective tool for a number of approaches in visual tracking, particularly for seeking a good balance between tracking accuracy and speed. However, correlation filter-based models are susceptible to wrong updates stemming from inaccurate tracking results. To date, very little effort has been devoted towards handling the correlation filter update problem. In this paper, we propose a novel approach to address the correlation filter update problem. In our approach, we update and maintain multiple correlation filter models in parallel, and we use deep reinforcement learning for the selection of an optimal correlation filter model among them. To facilitate the decision process in an efficient manner, we propose a decision-net to deal with target appearance modeling, which is trained through hundreds of challenging videos using proximal policy optimization and a lightweight learning network. An exhaustive evaluation of the proposed approach on the OTB100 and OTB2013 benchmarks shows that the approach is effective enough to achieve the average success rate of 62.3% and the average precision score of 81.2%, both exceeding the performance of traditional correlation filter-based trackers.

Chinese Shanshui is a landscape painting document mainly drawing mountain and water, which is popular in Chinese culture. However, it is very challenging to create this by general people. In this paper, we propose an interactive and generative approach to automatically generate the Chinese Shanshui painting documents based on users' input, where the users only need to sketch simple lines to represent their ideal landscape without any professional Shanshui painting skills. This sketch-to-Shanshui translation is optimized by the model of cycle Generative Adversarial Networks (GAN). To evaluate the proposed approach, we collected a large set of both sketch data and Chinese Shanshui painting data to train the model of cycle-GAN, and developed an interactive system called Shanshui-DaDA (i.e., Design and Draw with AI) to generate Chinese Shanshui painting documents in real-time. The experimental results show that this system can generate satisfied Chinese Shanshui painting documents by general users.

Mixture of factor analyzers is a fundamental model in unsupervised learning, which is particularly useful for high dimensional data. Recent efforts on deep auto-encoding mixture models made a fruitful progress in clustering. However, in most cases, their performance depends highly on the results of pre-training. Moreover, they tend to ignore the prior information when making clustering assignment, leading to a less strict inference and consequently limiting the performance. In this paper, we propose an end-to-end Bayesian nonparametric generalization of deep mixture model with a Variational Auto-Encoder (VAE) framework. Specifically, we develop a novel model called VSB-DVM exploiting the Variational Stick-Breaking Process to design a Deep Variational Mixture Model. Distinct from the existing deep auto-encoding mixture models, this novel unsupervised deep generative model can learn low-dimensional representations and clustering simultaneously without pre-training. Importantly, a strict inference is proposed using weights of stick-breaking process in a variational way. Furthermore, able to capture the richer statistical structure of the data, VSB-DVM can also generate highly realistic samples for any specified cluster. A series of experiments are carried out, both qualitatively and quantitatively, on benchmark clustering and generation tasks. Comparative results show that the proposed model is able to generate diverse and high-quality samples of data, and also achieves encouraging clustering results outperforming the state-of-the-art algorithms on four real-world datasets.

Adversarial examples, referred to as augmented data points generated by imperceptible perturbations of input samples, have recently drawn much attention. Well-crafted adversarial examples may even mislead state-of-the-art deep neural network (DNN) models to make wrong predictions easily. To alleviate this problem, many studies have focused on investigating how adversarial examples can be generated and/or effectively handled. All existing works tackle this problem in the Euclidean space. In this paper, we extend the learning of adversarial examples to the more general Riemannian space over DNNs. The proposed work is important in that (1) it is a generalized learning methodology since Riemmanian space will be degraded to the Euclidean space in a special case; (2) it is the first work to tackle the adversarial example problem tractably through the perspective of Riemannian geometry; (3) from the perspective of geometry, our method leads to the steepest direction of the loss function, by considering the second order information of the loss function. We also provide a theoretical study showing that our proposed method can truly find the descent direction for the loss function, with a comparable computational time against traditional adversarial methods. Finally, the proposed framework demonstrates superior performance over traditional counterpart methods, using benchmark data including MNIST, CIFAR-10 and SVHN.

Recent prevalent one stage detectors, such as single shot detector (SSD) and RetinaNet, are able to detect objects faster than two stage ones while maintaining comparable accuracy. To further boost the accuracy, many studies focus on enhancing the multi-scale feature pyramid. Most of these current proposals focus on strengthening features on one pyramid, ignoring the rich connection among different scale features. In contrast, we propose a novel multi-path design to fully utilize the localization and semantics information. First, we exploit the original SSD multi-scale features as our base pyramid. Then we fuse these features in different groups to generate multi-path feature pyramids. Finally, we combine these pyramids through a novel and effective aggregation module, to obtain the final informative pyramid for detection. Comparative experiments on benchmark PASCAL VOC and MS COCO datasets have shown that our proposed method outperforms many state-of-the-art detectors. As an illustrative example, for input image with size 512×512, we can achieve a mean Average Precision (mAP) of 81.8% on VOC2007 test and 33.1% mAP on COCO test-dev2015.

Conjugate gradient (CG) methods are a class of important methods for solving linear equations and nonlinear optimization problems. In this paper, we propose a new stochastic CG algorithm with variance reduction1and we prove its linear convergence with the Fletcher and Reeves method for strongly convex and smooth functions. We experimentally demonstrate that the CG with variance reduction algorithm converges faster than its counterparts for four learning models, which may be convex, nonconvex or nonsmooth. In addition, its area under the curve performance on six large-scale data sets is comparable to that of the LIBLINEAR solver for the L2-regularized L2-loss but with a significant improvement in computational efficiency.

Policy search in reinforcement learning (RL) is a practical approach to interact directly with environments in parameter spaces, that often deal with dilemmas of local optima and real-time sample collection. A promising algorithm, known as guided policy search (GPS), is capable of handling the challenge of training samples using trajectory-centric methods. It can also provide asymptotic local convergence guarantees. However, in its current form, the GPS algorithm cannot operate in sequential multitask learning scenarios. This is due to its batch-style training requirement, where all training samples are collectively provided at the start of the learning process. The algorithm’s adaptation is thus hindered for real-time applications, where training samples or tasks can arrive randomly. In this paper, the GPS approach is reformulated, by adapting a recently proposed, lifelong-learning method, and elastic weight consolidation. Specifically, Fisher information is incorporated to impart knowledge from previously learned tasks. The proposed algorithm, termed sequential multitask learning-GPS, is able to operate in sequential multitask learning settings and ensuring continuous policy learning, without catastrophic forgetting. Pendulum and robotic manipulation experiments demonstrate the new algorithms efficacy to learn control policies for handling sequentially arriving training samples, delivering comparable performance to the traditional, and batch-based GPS algorithm. In conclusion, the proposed algorithm is posited as a new benchmark for the real-time RL and robotics research community.

In this paper, we present a novel feature extraction technique, termed Three-Dimensional Local Energy-Based Shape Histogram (3D-LESH), and exploit it to detect breast cancer in volumetric medical images. The technique is incorporated as part of an intelligent expert system that can aid medical practitioners making diagnostic decisions. Analysis of volumetric images, slice by slice, is cumbersome and inefficient. Hence, 3D-LESH is designed to compute a histogram-based feature set from a local energy map, calculated using a phase congruency (PC) measure of volumetric Magnetic Resonance Imaging (MRI) scans in 3D space. 3D-LESH features are invariant to contrast intensity variations within different slices of the MRI scan and are thus suitable for medical image analysis. The contribution of this article is manifold. First, we formulate a novel 3D-LESH feature extraction technique for 3D medical images to analyse volumetric images. Further, the proposed 3D-LESH algorithmis, for the first time, applied to medical MRI images. The final contribution is the design of an intelligent clinical decision support system (CDSS) as a multi-stage approach, combining novel 3D-LESH feature extraction with machine learning classifiers, to detect cancer from breast MRI scans. The proposed system applies contrast-limited adaptive histogram equalisation (CLAHE) to the MRI images before extracting 3D-LESH features. Furthermore, a selected subset of these features is fed into a machine-learning classifier, namely, a support vector machine (SVM), an extreme learning machine (ELM) or an echo state network (ESN) classifier, to detect abnormalities and distinguish between different stages of abnormality. We demonstrate the performance of the proposed technique by its application to benchmark breast cancer MRI images. The results indicate high-performance accuracy of the proposed system (98%±0.0050, with an area under a receiver operating charactertistic curve value of 0.9900 ± 0.0050) with multiple classifiers. When compared with the state-of-the-art wavelet-based feature extraction technique, statistical analysis provides conclusive evidence of the significance of our proposed 3D-LESH algorithm.

Zero-shot learning (ZSL) aims at classifying examples for unseen classes (with no training examples) given some other seen classes (with training examples). Most existing approaches exploit intermedia-level information (e.g., attributes) to transfer knowledge from seen classes to unseen classes. A common practice is to first learn projections from samples to attributes on seen classes via a regression method, and then apply such projections to unseen classes directly. However, it turns out that such a manner of learning strategy easily causes projection domain shift problem and hubness problem, which hinder the performance of ZSL task. In this paper, we also formulate ZSL as an attribute regression problem. However, different from general regression-based solutions, the proposed approach is novel in three aspects. First, a class prototype rectification method is proposed to connect the unseen classes to the seen classes. Here, a class prototype refers to a vector representation of a class, and it is also known as a class center, class signature, or class exemplar. Second, an alternating learning scheme is proposed for jointly performing attribute regression and rectifying the class prototypes. Finally, a new objective function which takes into consideration both the attribute regression accuracy and the class prototype discrimination is proposed. By introducing such a solution, domain shift problem and hubness problem can be mitigated. Experimental results on three public datasets (i.e., CUB200-2011, SUN Attribute, and aPaY) well demonstrate the effectiveness of our approach.

Adversarial examples are augmented data points generated by imperceptible perturbation of input samples. They have recently drawn much attention with the machine learning and data mining community. Being difficult to distinguish from real examples, such adversarial examples could change the prediction of many of the best learning models including the state-of-the-art deep learning models. Recent attempts have been made to build robust models that take into account adversarial examples. However, these methods can either lead to performance drops or lack mathematical motivations. In this paper, we propose a unified framework to build robust machine learning models against adversarial examples. More specifically, using the unified framework, we develop a family of gradient regularization methods that effectively penalize the gradient of loss function w.r.t. inputs. Our proposed framework is appealing in that it offers a unified view to deal with adversarial examples. It incorporates another recently-proposed perturbation based approach as a special case. In addition, we present some visual effects that reveals semantic meaning in those perturbations, and thus support our regularization method and provide another explanation for generalizability of adversarial examples. By applying this technique to Maxout networks, we conduct a series of experiments and achieve encouraging results on two benchmark datasets. In particular, we attain the best accuracy on MNIST data (without data augmentation) and competitive performance on CIFAR-10 data.

Dimensionality Reduction (DR) is a fundamental yet active research topic in pattern recognition and machine learning. When used in classification, previous research usually performs DR separately, and then inputs the reduced features to other available models, e.g., Gaussian Mixture Model (GMM). Such independent learning could however significantly limit the classification performance, since the optimal subspace given by a particular DR approach may not be appropriate for the following classification model. More seriously, for high-dimensional data classification in the face of a limited number of samples (called small sample size or S3 problem), independent learning of DR and classification model may even deteriorate the classification accuracy. To solve this problem, we propose a joint learning model, called Two-layer Mixture of Factor Analyzers with Joint Factor Loading (2L-MJFA) for classification. More specifically, our proposed model enjoys a two-layer mixture structure, or a mixture of mixtures structure, with each component (representing each specific class) as another mixture model of Factor Analyzer (MFA). Importantly, all the involved factor analyzers are intentionally designed to share the same loading matrix. On one hand, such joint loading matrix can be considered as the dimensionality reduction matrix; on the other hand, a joint common matrix would largely reduce the parameters, making the proposed algorithm very suitable for S3 problems. We describe our model definition and propose a modified EM algorithm to optimize the model. A series of experiments demonstrates that our proposed model significantly outperforms the other three competitive algorithms on five data sets.

The SDA workshop at WSDM 2015 is the fifth International Workshop on Scalable Data Analytics, following the previous four workshops of SDA respectively held at IEEE Big Data 2013, PAKDD 2014, IEEE Big Data 2014, and IEEE ICDM 2014. This series of workshops aims to provide professionals, researchers, and technologists with a single forum where they can discuss and share the state-of-the-art theories and applications of scalable data analytics technologies. In particular, in the era of information explosion, the scientific, biomedical, and engineering research communities are undergoing a profound transformation where discoveries and innovations increasingly rely on massive amounts of data. The characteristics of volume, velocity, variety and veracity originated in the massive big data then bring challenges to current data analytics techniques. The focus of the fifth SDA is to discuss how we can scale up data analytics techniques for modeling and analyzing big data from various domains.

Despite the great success of graph-based transductive learning methods, most of them have serious problems in scalability and robustness. In this paper, we propose an efficient and robust graph-based transductive classification method, called minimum tree cut (MTC), which is suitable for large-scale data. Motivated from the sparse representation of graph, we approximate a graph by a spanning tree. Exploiting the simple structure, we develop a linear-time algorithm to label the tree such that the cut size of the tree is minimized. This significantly improves graph-based methods, which typically have a polynomial time complexity. Moreover, we theoretically and empirically show that the performance of MTC is robust to the graph construction, overcoming another big problem of traditional graph-based methods. Extensive experiments on public data sets and applications on web-spam detection and interactive image segmentation demonstrate our method's advantages in aspect of accuracy, speed, and robustness.

Text detection in natural scene images is an important prerequisite for many content-based image analysis tasks. In this paper, we propose an accurate and robust method for detecting texts in natural scene images. A fast and effective pruning algorithm is designed to extract Maximally Stable Extremal Regions (MSERs) as character candidates using the strategy of minimizing regularized variations. Character candidates are grouped into text candidates by the single-link clustering algorithm, where distance weights and clustering threshold are learned automatically by a novel self-training distance metric learning algorithm. The posterior probabilities of text candidates corresponding to non-text are estimated with a character classifier; text candidates with high non-text probabilities are eliminated and texts are identified with a text classifier. The proposed system is evaluated on the ICDAR 2011 Robust Reading Competition database; the f-measure is over 76%, much better than the state-of-the-art performance of 71%. Experiments on multilingual, street view, multi-orientation and even born-digital databases also demonstrate the effectiveness of the proposed method.

This letter considers the combination of multiple classification and clustering results to improve the prediction accuracy. First, an object-similarity graph is constructed from multiple clustering results. The labels predicted by the classification models are then propagated on this graph to adaptively satisfy the smoothness of the prediction over the graph. The convex learning problem is efficiently solved by the label propagation algorithm. A semi-supervised extension is also provided to further improve the performance. Experiments on 11 tasks identify the validity of the proposed models.

Machine learning based on graph representation, or manifold learning, has attracted great interest in recent years. As the discrete approximation of data manifold, the graph plays a crucial role in these kinds of learning approaches. In this paper, we propose a novel learning method for graph construction, which is distinct from previous methods in that it solves an optimization problem with the aim of directly preserving the local information of the original data set. We show that the proposed objective has close connections with the popular Laplacian Eigenmap problem, and is hence well justified. The optimization turns out to be a quadratic programming problem with n(n-1)/2 variables (n is the number of data points). Exploiting the sparsity of the graph, we further propose a more efficient cutting plane algorithm to solve the problem, making the method better scalable in practice. In the context of clustering and semi-supervised learning, we demonstrated the advantages of our proposed method by experiments.

The well-known feature transformation model of Fisher linear discriminant analysis (FDA) can be decomposed into an equivalent two-step approach: whitening followed by principal component analysis (PCA) in the whitened space. By proving that whitening is the optimal linear transformation to the Euclidean space in the sense of minimum log-determinant divergence, we propose a transformation model called class conditional decor relation (CCD). The objective of CCD is to diagonalize the covariance matrices of different classes simultaneously, which is efficiently optimized using a modified Jacobi method. CCD is effective to find the common principal components among multiple classes. After CCD, the variables become class conditionally uncorrelated, which will benefit the subsequent classification tasks. Combining CCD with the nearest class mean (NCM) classification model can significantly improve the classification accuracy. Experiments on 15 small-scale datasets and one large-scale dataset (with 3755 classes) demonstrate the scalability of CCD for different applications. We also discuss the potential applications of CCD for other problems such as Gaussian mixture models and classifier ensemble learning.

We propose and implement a robust text detection system, which is a prominent step-in for text retrieval in natural scene images or videos. Our system includes several key components: (1) A fast and effective pruning algorithm is designed to extract Maximally Stable Extremal Regions as character candidates using the strategy of minimizing regularized variations. (2) Character candidates are grouped into text candidates by the single-link clustering algorithm, where distance weights and threshold of clustering are learned automatically by a novel self-training distance metric learning algorithm. (3) The posterior probabilities of text candidates corresponding to non-text are estimated with an character classifier; text candidates with high probabilities are then eliminated and finally texts are identified with a text classifier. The proposed system is evaluated on the ICDAR 2011 Robust Reading Competition dataset and a publicly available multilingual dataset; the f measures are over 76% and 74% which are significantly better than the state-of-the-art performances of 71% and 65%, respectively.

In this paper, we propose an efficient and robust algorithm for graph-based transductive classification. After approximating a graph with a spanning tree, we develop a linear-time algorithm to label the tree such that the cut size of the tree is minimized. This significantly improves typical graph-based methods, which either have a cubic time complexity (for a dense graph) or O(kn2) (for a sparse graph with k denoting the node degree). Furthermore, our method shows great robustness to the graph construction both theoretically and empirically; this overcomes another big problem of traditional graph-based methods. In addition to its good scalability and robustness, the proposed algorithm demonstrates high accuracy. In particular, on a graph with 400,000 nodes (in which 10,000 nodes are labeled) and 10,455,545 edges, our algorithm achieves the highest accuracy of 99.6% but takes less than 10 seconds to label all the unlabeled data.

In this paper, we present a semi-supervised method to learn a low rank Mahalanobis distance function. Based on an approximation to the projection distance from a manifold, we propose a novel parametric manifold regularizer. In contrast to previous approaches that usually exploit side information only, our proposed method can further take advantages of the intrinsic manifold information from data. In addition, we focus on learning a metric of low rank directly, this is different from traditional approaches that often enforce the l1norm on the metric. The resulting configuration is convex with respect to the manifold structure and the distance function, respectively. We solve it with an alternating optimization algorithm, which proves effective to find a satisfactory solution. For efficient implementation, we even present a fast algorithm, in which the manifold structure and the distance function are learned independently without alternating minimization. Experimental results over 12 standard UCI data sets demonstrate the advantages of our method.

Dimension reduction has been widely used in real-world applications such as image retrieval and document classification. In many scenarios, different features (or multiview data) can be obtained, and how to duly utilize them is a challenge. It is not appropriate for the conventional concatenating strategy to arrange features of different views into a long vector. That is because each view has its specific statistical property and physical interpretation. Even worse, the performance of the concatenating strategy will deteriorate if some views are corrupted by noise. In this paper, we propose a multiview stochastic neighbor embedding (m-SNE) that systematically integrates heterogeneous features into a unified representation for subsequent processing based on a probabilistic framework. Compared with conventional strategies, our approach can automatically learn a combination coefficient for each view adapted to its contribution to the data embedding. This combination coefficient plays an important role in utilizing the complementary information in multiview data. Also, our algorithm for learning the combination coefficient converges at a rate of O(1/k(2)), which is the optimal rate for smooth problems. Experiments on synthetic and real data sets suggest the effectiveness and robustness of m-SNE for data visualization, image retrieval, object categorization, and scene recognition.

We consider the problem of similar Chinese character recognition in this paper. Engaging the Average Symmetric Uncertainty (ASU) criterion to measure the correlation between different image regions and the class label, we manage to detect the most critical regions for each pair of similar characters. These critical regions are proved to contain more discriminative information and hence can largely benefit the classification accuracy for similar characters. We conduct a series of experiments on the CASIA Chinese character data set. Experimental results show that our proposed method is superior to three competitive approaches in terms of both accuracy and efficiency.

In this paper, we propose a novel discriminant analysis method, called Minimal Distance Maximization (MDM). In contrast to the traditional LDA, which actually maximizes the average divergence among classes, MDM attempts to find a low-dimensional subspace that maximizes the minimal (worst-case) divergence among classes. This ``minimal" setting solves the problem caused by the ``average" setting of LDA that tends to merge similar classes with smaller divergence when used for multi-class data. Furthermore, we elegantly formulate the worst-case problem as a convex problem, making the algorithm solvable for larger data sets. Experimental results demonstrate the advantages of our proposed method against five other competitive approaches on one synthetic and six real-life data sets.

There has been significant recent interest in sparse metric learning (SML) in which we simultaneously learn both a good distance metric and a low-dimensional representation. Unfortunately, the performance of existing sparse metric learning approaches is usually limited because the authors assumed certain problem relaxations or they target the SML objective indirectly. In this paper, we propose a Generalized Sparse Metric Learning method (GSML). This novel framework offers a unified view for understanding many of the popular sparse metric learning algorithms including the Sparse Metric Learning framework proposed, the Large Margin Nearest Neighbor (LMNN), and the D-ranking Vector Machine (D-ranking VM). Moreover, GSML also establishes a close relationship with the Pairwise Support Vector Machine. Furthermore, the proposed framework is capable of extending many current non-sparse metric learning models such as Relevant Vector Machine (RCA) and a state-of-the-art method proposed into their sparse versions. We present the detailed framework, provide theoretical justifications, build various connections with other models, and propose a practical iterative optimization method, making the framework both theoretically important and practically scalable for medium or large datasets. A series of experiments show that the proposed approach can outperform previous methods in terms of both test accuracy and dimension reduction, on six real-world benchmark datasets.

We consider the task of Self-taught Learning (STL) from unlabeled data. In contrast to semi-supervised learning, which requires unlabeled data to have the same set of class labels as labeled data, STL can transfer knowledge from different types of unlabeled data. STL uses a three-step strategy: (1) learning high-level representations from unlabeled data only, (2) re-constructing the labeled data via such representations and (3) building a classifier over the re-constructed labeled data. However, the high-level representations which are exclusively determined by the unlabeled data, may be inappropriate or even misleading for the latter classifier training process. In this paper, we propose a novel Supervised Self-taught Learning (SSTL) framework that successfully integrates the three isolated steps of STL into a single optimization problem. Benefiting from the interaction between the classifier optimization and the process of choosing high-level representations, the proposed model is able to select those discriminative representations which are more appropriate for classification. One important feature of our novel framework is that the final optimization can be iteratively solved with convergence guaranteed. We evaluate our novel framework on various data sets. The experimental results show that the proposed SSTL can outperform STL and traditional supervised learning methods in certain instances.

In this paper we study the problem of learning a low-dimensional (sparse) distance matrix. We propose a novel metric learning model which can simultaneously conduct dimension reduction and learn a distance matrix. The sparse representation involves a mixed-norm regularization which is non-convex. We then show that it can be equivalently formulated as a convex saddle (min-max) problem. From this saddle representation, we develop an efficient smooth optimization approach for sparse metric learning although the learning model is based on a non-differential loss function. This smooth optimization approach has an optimal convergence rate of $O(1 /\ell^2)$ for smooth problems where $\ell$ is the iteration number. Finally, we run experiments to validate the effectiveness and efficiency of our sparse metric learning model on various datasets.

BACKGROUND: Protein fold recognition is a key step in protein three-dimensional (3D) structure discovery. There are multiple fold discriminatory data sources which use physicochemical and structural properties as well as further data sources derived from local sequence alignments. This raises the issue of finding the most efficient method for combining these different informative data sources and exploring their relative significance for protein fold classification. Kernel methods have been extensively used for biological data analysis. They can incorporate separate fold discriminatory features into kernel matrices which encode the similarity between samples in their respective data sources. RESULTS: In this paper we consider the problem of integrating multiple data sources using a kernel-based approach. We propose a novel information-theoretic approach based on a Kullback-Leibler (KL) divergence between the output kernel matrix and the input kernel matrix so as to integrate heterogeneous data sources. One of the most appealing properties of this approach is that it can easily cope with multi-class classification and multi-task learning by an appropriate choice of the output kernel matrix. Based on the position of the output and input kernel matrices in the KL-divergence objective, there are two formulations which we respectively refer to as MKLdiv-dc and MKLdiv-conv. We propose to efficiently solve MKLdiv-dc by a difference of convex (DC) programming method and MKLdiv-conv by a projected gradient descent algorithm. The effectiveness of the proposed approaches is evaluated on a benchmark dataset for protein fold recognition and a yeast protein function prediction problem. CONCLUSION: Our proposed methods MKLdiv-dc and MKLdiv-conv are able to achieve state-of-the-art performance on the SCOP PDB-40D benchmark dataset for protein fold prediction and provide useful insights into the relative significance of informative data sources. In particular, MKLdiv-dc further improves the fold discrimination accuracy to 75.19% which is a more than 5% improvement over competitive Bayesian probabilistic and SVM margin-based kernel learning methods. Furthermore, we report a competitive performance on the yeast protein function prediction problem.

Support vector machines (SVM) are state-of-the-art classifiers. Typically L2-norm or L1-norm is adopted as a regularization term in SVMs, while other norm-based SVMs, for example, the L0-norm SVM or even the L(infinity)-norm SVM, are rarely seen in the literature. The major reason is that L0-norm describes a discontinuous and nonconvex term, leading to a combinatorially NP-hard optimization problem. In this letter, motivated by Bayesian learning, we propose a novel framework that can implement arbitrary norm-based SVMs in polynomial time. One significant feature of this framework is that only a sequence of sequential minimal optimization problems needs to be solved, thus making it practical in many real applications. The proposed framework is important in the sense that Bayesian priors can be efficiently plugged into most learning methods without knowing the explicit form. Hence, this builds a connection between Bayesian learning and the kernel machines. We derive the theoretical framework, demonstrate how our approach works on the L0-norm SVM as a typical example, and perform a series of experiments to validate its advantages. Experimental results on nine benchmark data sets are very encouraging. The implemented L0-norm is competitive with or even better than the standard L2-norm SVM in terms of accuracy but with a reduced number of support vectors, -9.46% of the number on average. When compared with another sparse model, the relevance vector machine, our proposed algorithm also demonstrates better sparse properties with a training speed over seven times faster.

In automated text categorization, given a small number of labeled documents, it is very challenging, if not impossible, to build a reliable classifier that is able to achieve high classification accuracy. To address this problem, a novel web-assisted text categorization framework is proposed in this paper. Important keywords are first automatically identified from the available labeled documents to form the queries. Search engines are then utilized to retrieve from the Web a multitude of relevant documents, which are then exploited by a semi-supervised framework. To our best knowledge, this work is the first study of this kind. Extensive experimental study shows the encouraging results of the proposed text categorization framework: using Google as the web search engine, the proposed framework is able to reduce the classification error by 30% when compared with the state-of-the-art supervised text categorization method.

Zero-norm, defined as the number of non-zero elements in a vector, is an ideal quantity for feature selection. However, minimization of zero-norm is generally regarded as a combinatorially difficult optimization problem. In contrast to previous methods that usually optimize a surrogate of zero-norm, we propose a direct optimization method to achieve zero-norm for feature selection in this paper. Based on Expectation Maximization (EM), this method boils down to solving a sequence of Quadratic Programming problems and hence can be practically optimized in polynomial time. We show that the proposed optimization technique has a nice Bayesian interpretation and converges to the true zero norm asymptotically, provided that a good starting point is given. Following the scheme of our proposed zero-norm, we even show that an arbitrary-norm based Support Vector Machine can be achieved in polynomial time. A series of experiments demonstrate that our proposed EM based zero-norm outperforms other state-of-the-art methods for feature selection on biological microarray data and UCI data, in terms of both the accuracy and the learning efficiency.

We consider the problem of semi-supervised learning (SSL) from general unlabeled data, which may contain irrelevant samples. Within the binary setting, our model manages to better utilize the information from unlabeled data by formulating them as a three-class (-1,+1, 0) mixture, where class 0 represents the irrelevant data. This distinguishes our work from the traditional SSL problem where unlabeled data are assumed to contain relevant samples only, either +1 or -1, which are forced to be the same as the given labeled samples. This work is also different from another family of popular models, universum learning (universum means "irrelevant" data), in that the universum need not to be specified beforehand. One significant contribution of our proposed framework is that such irrelevant samples can be automatically detected from the available unlabeled data, even though they are mixed with relevant data. This hence presents a general SSL framework that does not force "clean" unlabeled data.More importantly, we formulate this general learning framework as a Semi-definite Programming problem, making it solvable in polynomial time. A series of experiments demonstrate that the proposed framework can outperform the traditional SSL on both synthetic and real data.

Minimax Probability Machine (MPM), learning a decision function by minimizing the maximum probability of misclassification, has demonstrated very promising performance in classification and regression. However, MPM is often challenged for its slow training and test procedures. Aiming to solve this problem, we propose an efficient model named Minimax Clustering Probability Machine (MCPM). Following many traditional methods, we represent training data points by several clusters. Different from these methods, a Generalized Probability Product Kernel is appropriately defined to grasp the inner distributional information over the clusters. Incorporating clustering information via a non-linear kernel, MCPM can fast train and test in classification problem with promising performance. Another appealing property of the proposed approach is that MCPM can still derive an explicit worst-case accuracy bound for the decision boundary. Experimental results on synthetic and real data validate the effectiveness of MCPM for classification while attaining high accuracy.

In this paper, we propose a novel large margin classifier, called the maxi-min margin machine M(4). This model learns the decision boundary both locally and globally. In comparison, other large margin classifiers construct separating hyperplanes only either locally or globally. For example, a state-of-the-art large margin classifier, the support vector machine (SVM), considers data only locally, while another significant model, the minimax probability machine (MPM), focuses on building the decision hyperplane exclusively based on the global information. As a major contribution, we show that SVM yields the same solution as M(4) when data satisfy certain conditions, and MPM can be regarded as a relaxation model of M(4). Moreover, based on our proposed local and global view of data, another popular model, the linear discriminant analysis, can easily be interpreted and extended as well. We describe the M(4) model definition, provide a geometrical interpretation, present theoretical justifications, and propose a practical sequential conic programming method to solve the optimization problem. We also show how to exploit Mercer kernels to extend M(4) for nonlinear classifications. Furthermore, we perform a series of evaluations on both synthetic data sets and real-world benchmark data sets. Comparison with SVM and MPM demonstrates the advantages of our new model.

Handwritten numeral recognition is an important pattern recognition task. It can be widely used in various domains, e.g., bank money recognition, which requires a very high recognition rate. As a state-of-the-art classifier, support vector machine (SVM), has been extensively used in this area. Typically, SVM is trained in a batch model, i.e., all data points are simultaneously input for training the classification boundary. However, some slightly exceptional data, only accounting for a small proportion, are critical for the recognition rates. Training a classifier among all the data may possibly treat such legal but slightly exceptional samples as "noise ". In this paper, we propose a novel approach to attack this problem. This approach exploits a two-stage framework by using difference features. In the first stage, a regular SVM is trained on all the training data; in the second stage, only the samples misclassified in the first stage are specially considered. Therefore, the performance can be lifted. The number of misclassifications is often small because of the good performance of SVM. This will present difficulties in training an accurate SVM engine only for these misclassified samples. We then further propose a multi-way to binary approach using difference features. This approach successfully transforms multi-category classification to binary classification and expands the training samples greatly. In order to evaluate the proposed method, experiments are performed on 10,000 handwritten numeral samples extracted from real banks forms. This new algorithm achieves 99.0% accuracy. In comparison, the traditional SVM only gets 98.4%.

Character degradation is a big problem for machine printed character recognition. Two main reasons for degradation are extrinsic image degradation such as blurring and low image dimension, and intrinsic degradation caused by font variations. A recognition method that combines two complementary classifiers is proposed in this paper. The local feature based classifier extracts the local contour direction changes, which is effective for character patterns with less structure deterioration. The global feature based classifier extracts the texture distribution of the character image, which is effective when the character structure is hard to discriminate. The two complementary classifiers are combined by candidate fusion in a coarse-to-fine style. Experiments are carried on degraded Chinese character recognition. The results prove the effectiveness of our method.

We consider the regression problem for financial time series. Typically, financial time series are non-stationary and volatile in nature. Because of its good generalization power and the tractability of the problem, the Support Vector Regression (SVR) has been extensively applied in financial time series prediction. The standard SVR adopts the lp-norm (p = 1 or 2) to model the functional complexity of the whole data set and employs a fixed ε-tube to tolerate noise. Although this approach has proved successful both theoretically and empirically, it considers data in a global fashion only. Therefore it may lack the flexibility to capture the local trend of data; this is a critical aspect of volatile data, especially financial time series data. Aiming to address this issue, we propose the Local Support Vector Regression (LSVR) model. This novel model is demonstrated to provide a systematic and automatic scheme to adapt the margin locally and flexibly; the margin is fixed globally in the standard SVR. Therefore, the LSVR can tolerate noise adaptively. We provide both theoretical justifications and empirical evaluations for this novel model. The experimental results on synthetic data and real financial data demonstrate its advantages over the standard SVR.

Imbalanced learning is a challenged task in machine learning. In this context, the data associated with one class are far fewer than those associated with the other class. Traditional machine learning methods seeking classification accuracy over a full range of instances are not suitable to deal with this problem, since they tend to classify all the data into a majority class, usually the less important class. In this correspondence, the authors describe a new approach named the biased minimax probability machine (BMPM) to deal with the problem of imbalanced learning. This BMPM model is demonstrated to provide an elegant and systematic way for imbalanced learning. More specifically, by controlling the accuracy of the majority class under all possible choices of class-conditional densities with a given mean and covariance matrix, this model can quantitatively and systematically incorporate a bias for the minority class. By establishing an explicit connection between the classification accuracy and the bias, this approach distinguishes itself from the many current imbalanced-learning methods; these methods often impose a certain bias on the minority data by adapting intermediate factors via the trial-and-error procedure. The authors detail the theoretical foundation, prove its solvability, propose an efficient optimization algorithm, and perform a series of experiments to evaluate the novel model. The comparison with other competitive methods demonstrates the effectiveness of this new model.

A new large margin classifier, named Maxi-Min Margin Machine (M4) is proposed in this paper. This new classifier is constructed based on both a "local: and a "global" view of data, while the most popular large margin classifier, Support Vector Machine (SVM) and the recently-proposed important model, Minimax Probability Machine (MPM) consider data only either locally or globally. This new model is theoretically important in the sense that SVM and MPM can both be considered as its special case. Furthermore, the optimization of M4 can be cast as a sequential conic programming problem, which can be solved efficiently. We describe the M4 model definition, provide a clear geometrical interpretation, present theoretical justifications, propose efficient solving methods, and perform a series of evaluations on both synthetic data sets and real world benchmark data sets. Its comparison with SVM and MPM also demonstrates the advantages of our new model.

Recently, support vector machines (SVMs) have been engaged on relevance feedback tasks in content-based image retrieval. Typical approaches by SVMs treat the relevance feedback as a strict binary classification problem. However, these approaches do not consider an important issue of relevance feedback, i.e. the unbalanced dataset problem, in which the negative instances largely outnumber the positive instances. For solving this problem, we propose a novel technique to formulate the relevance feedback based on a modified SVM called biased support vector machine (Biased SVM or BSVM). Mathematical formulation and explanations are provided for showing the advantages. Experiments are conducted to evaluate the performance of our algorithms, in which promising results demonstrate the effectiveness of our techniques.

Discriminative classifiers such as support vector machines directly learn a discriminant function or a posterior probability model to perform classification. On the other hand, generative classifiers often learn a joint probability model and then use Bayes rules to construct a posterior classifier from this model. In general, generative classifiers are not as accurate as discriminant classifier. However generative classifiers provide a principled way to handle the missing information problems, which discriminant classifiers cannot easily deal with. To achieve good performances in various classification tasks, it is better to combine these two strategies. In this paper, we develop a novel method to iteratively train a kind of generative Bayesian classifier: Bayesian Chow-Liu multinet classifier in a discriminative way. Different with the traditional Bayesian multinet classifiers, our discriminative method adds into the optimization function a penalty item, which represents the divergence between classes as big as possible. We state the theoretical justification, outline of the algorithm and also perform a series of experiments to demonstrate the advantages of our method. The experiments results are promising and encouraging.