Table images present unique challenges for effective and efficient understanding due to the need for question-specific focus and the presence of redundant background regions. Existing Multimodal Large Language Model (MLLM) approaches often overlook these characteristics, resulting in uninformative and redundant visual representations. To address these issues, we aim to generate visual features that are both informative and compact for improved table understanding. We first propose progressive question conditioning, which injects the question into Vision Transformer layers with gradually increasing frequency, considering each layer’s capacity to handle additional information, to generate question-aware visual features. To reduce redundancy, we introduce a pruning strategy that discards background tokens, thereby improving efficiency. To mitigate information loss from pruning, we further propose token focusing, a training strategy that encourages the model to concentrate essential information in the retained tokens. By combining these approaches, we present TabFlash, an efficient and effective MLLM for table understanding. TabFlash achieves state-of-the-art performance, outperforming both open-source and proprietary MLLMs, while requiring 27% less FLOPs and 30% less memory usage compared to the second-best MLLM.

Large language models (LLMs) have demonstrated their instruction-following capabilities and achieved powerful performance on various tasks. Inspired by their success, recent works in the molecular domain have led to the development of large molecular language models (LMLMs) that integrate 1D molecular strings or 2D molecular graphs into the language models. However, existing LMLMs often suffer from hallucination and limited robustness, largely due to inadequate integration of diverse molecular modalities such as 1D sequences, 2D molecular graphs, and 3D conformations. To address these limitations, we propose CoLLaMo, a large language model-based molecular assistant equipped with a multi-level molecular modality-collaborative projector. The relation-aware modality-collaborative attention mechanism in the projector facilitates fine-grained and relation-guided information exchange between atoms by incorporating 2D structural and 3D spatial relations. Furthermore, we present a molecule-centric new automatic measurement, including a hallucination assessment metric and GPT-based caption quality evaluation to address the limitations of token-based generic evaluation metrics (i.e., BLEU) widely used in assessing molecular comprehension of LMLMs. Our extensive experiments demonstrate that our CoLLaMo enhances the molecular modality generalization capabilities of LMLMs, achieving the best performance on multiple tasks, including molecule captioning, computed property QA, descriptive property QA, motif counting, and IUPAC name prediction.

Vision-Language Models (VLMs) have been widely used in various visual recognition tasks due to their remarkable generalization capabilities. As these models grow in size and complexity, fine-tuning becomes costly, emphasizing the need to reuse adaptation knowledge from 'weaker' models to efficiently enhance 'stronger' ones. However, existing adaptation transfer methods exhibit limited transferability across models due to their model-specific design and high computational demands. To tackle this, we propose Transferable Model-agnostic adapter (TransMiter), a light-weight adapter that improves vision-language models 'without backpropagation'. TransMiter captures the knowledge gap between pre-trained and fine-tuned VLMs, in an 'unsupervised' manner. Once trained, this knowledge can be seamlessly transferred across different models without the need for backpropagation. Moreover, TransMiter consists of only a few layers, inducing a negligible additional inference cost. Notably, supplementing the process with a few labeled data further yields additional performance gain, often surpassing a fine-tuned stronger model, with a marginal training cost. Experimental results and analyses demonstrate that TransMiter effectively and efficiently transfers adaptation knowledge while preserving generalization abilities across VLMs of different sizes and architectures in visual recognition tasks.

Despite the remarkable capabilities of Large Vision Language Models (LVLMs), they still lack detailed knowledge about specific entities. Retrieval-augmented Generation (RAG) is a widely adopted solution that enhances LVLMs by providing additional contexts from an external Knowledge Base. However, we observe that previous decoding methods for RAG are sub-optimal as they fail to sufficiently leverage multiple relevant contexts and suppress the negative effects of irrelevant contexts. To this end, we propose Relevance-aware Multi-context Contrastive Decoding (RMCD), a novel decoding method for RAG. RMCD outputs a final prediction by combining outputs predicted with each context, where each output is weighted based on its relevance to the question. By doing so, RMCD effectively aggregates useful information from multiple relevant contexts while also counteracting the negative effects of irrelevant ones. Experiments show that RMCD consistently outperforms other decoding methods across multiple LVLMs, achieving the best performance on three knowledge-intensive visual question-answering benchmarks. Also, RMCD can be simply applied by replacing the decoding method of LVLMs without additional training. Analyses also show that RMCD is robust to the retrieval results, consistently performing the best across the weakest to the strongest retrieval results.

Abstract Pre-trained vision-language models ( e.g ., CLIP) have shown impressive success in various computer vision tasks with their generalization capability. Recently, parameter-efficient fine-tuning (PEFT) approaches have been actively explored to effectively and efficiently adapt the pre-trained vision-language models to a variety of downstream tasks. However, most existing PEFT approaches suffer from a task overfitting issue since the general knowledge of the pre-trained models is forgotten while a small number of learnable parameters in soft prompts/adapters are fine-tuned on a small data set from a specific target task. Thus, we propose a P arameter- E fficient F ine- T uning via Meta - R egularization (PEFT-MetaR) to improve the generalizability of parameter-efficient fine-tuning methods for vision-language models. Specifically, PEFT-MetaR meta-learns both the regularizer and learnable parameters to harness the task-specific knowledge from the downstream tasks and task-agnostic general knowledge from the pretrained models. Further, PEFT-MetaR augments the task to generate multiple virtual tasks to alleviate the meta-overfitting. In addition, we provide the analysis to comprehend how PEFT-MetaR improves the generalizability from the perspective of the gradient alignment. Our experiments demonstrate that PEFT-MetaR improves the generalizability of parameter-efficient fine-tuning methods on various datasets.

Visual Relationship Detection (VRD) has significantly advanced with Transformer-based architectures. However, we identify two fundamental drawbacks in conventional label assignment methods used for training Transformer-based VRD models, where ground-truth (GT) annotations are matched to model predictions. In conventional assignment, queries are trained to detect all relations rather than specializing in specific ones, resulting in ‘unspecialized’ queries. Also, each ground-truth (GT) annotation is assigned to only one prediction under conventional assignment, suppressing other near-correct predictions by labeling them as ‘no relation’. To address these issues, we introduce a novel method called Groupwise Query Spe ci a lization and Q uality-Aware Multi-Assignment (SpeaQ). Groupwise Query Specialization clusters queries and relations into exclusive groups, promoting specialization by assigning a set of relations only to a corresponding query group. Quality-Aware Multi-Assignment enhances training signals by allowing multiple predictions closely matching the GT to be positively assigned. Additionally, we introduce dynamic query reallocation, which transfers queries from high- to low-performing groups for balanced training. Experimental results demonstrate that SpeaQ+, combining SpeaQ with dynamic query reallocation, consistently improves performance across seven baseline models on five benchmarks without additional inference cost.

Attribute classification is crucial for identifying specific characteristics within image regions. Vision-Language Models (VLMs) have been effective in zero-shot tasks by leveraging their general knowledge from large-scale datasets. Recent studies demonstrate that transformer-based models with class-wise queries can effectively address zero-shot multi-label classification. However, poor utilization of the relationship between seen and unseen attributes makes the model lack generalizability. Additionally, attribute classification generally involves many attributes, making maintaining the model’s scalability difficult. To address these issues, we propose Super-class guided transFormer (SugaFormer), a novel framework that leverages super-classes to enhance scalability and generalizability for zero-shot attribute classification. SugaFormer employs Super-class Query Initialization (SQI) to reduce the number of queries, utilizing common semantic information from super-classes, and incorporates Multi-context Decoding (MD) to handle diverse visual cues. To strengthen generalizability, we introduce two knowledge transfer strategies that utilize VLMs. During training, Super-class guided Consistency Regularization (SCR) aligns model’s features with VLMs using super-class guided prompts, and during inference, Zero-shot Retrieval-based Score Enhancement (ZRSE) refines predictions for unseen attributes. Extensive experiments demonstrate that SugaFormer achieves state-of-the-art performance across three widely-used attribute classification benchmarks under zero-shot, and cross-dataset transfer settings.

Despite the advancements of Video Large Language Models (VideoLLMs) in various tasks, they struggle with fine-grained temporal understanding, such as Dense Video Captioning (DVC). DVC is a complicated task of describing all events within a video while also temporally localizing them, which integrates multiple fine-grained tasks, including video segmentation, video captioning, and temporal video grounding. Previous VideoLLMs attempt to solve DVC in a single step, failing to utilize their reasoning capability. Moreover, previous training objectives for VideoLLMs do not fully reflect the evaluation metrics, therefore not providing supervision directly aligned to target tasks. To address such a problem, we propose a novel framework named VidChain comprised of Chain-of-Tasks (CoTasks) and Metric-based Direct Preference Optimization (M-DPO). CoTasks decompose a complex task into a sequence of sub-tasks, allowing VideoLLMs to leverage their reasoning capabilities more effectively. M-DPO aligns a VideoLLM with evaluation metrics, providing fine-grained supervision to each task that is well-aligned with metrics. Applied to two different VideoLLMs, VidChain consistently improves their fine-grained video understanding, thereby outperforming previous VideoLLMs on two different DVC benchmarks and also on the temporal video grounding task.

For the deployment of neural networks in resource-constrained environments, prior works have built lightweight architectures with convolution and attention for capturing local and global dependencies, respectively. Recently, the state space model (SSM) has emerged as an effective operation for global interaction with its favorable linear computational cost in the number of tokens. To harness the efficacy of SSM, we introduce Efficient Vision Mamba (EfficientViM), a novel architecture built on hidden state mixer-based state space duality (HSM-SSD) that efficiently captures global dependencies with further reduced computational cost. With the observation that the runtime of the SSD layer is driven by the linear projections on the input sequences, we redesign the original SSD layer to perform the channel mixing operation within compressed hidden states in the HSM-SSD layer. Additionally, we propose multi-stage hidden state fusion to reinforce the representation power of hidden states and provide the design to alleviate the bottleneck caused by the memory-bound operations. As a result, the EfficientViM family achieves a new state-of-the-art speed-accuracy trade-off on ImageNet-1k, offering up to a 0.7% performance improvement over the second-best model SHViT with faster speed. Further, we observe significant improvements in throughput and accuracy compared to prior works, when scaling images or employing distillation training. Code is available at https://github.com/mlvlab/EfficientViM.

Transformers have demonstrated remarkable success across vision, language, and video. Yet, increasing task complexity has led to larger models and more tokens, raising the quadratic cost of self-attention and the overhead of GPU memory access. To reduce the computation cost of self-attention, prior work has proposed token compression techniques that drop redundant or less informative tokens. Meanwhile, fused attention kernels such as FlashAttention have been developed to alleviate memory overhead by avoiding attention map construction and its associated I/O to HBM. This, however, makes it incompatible with most training-free token compression methods, which rely on attention maps to determine token importance. Here, we propose Representation Shift, a training-free, model-agnostic metric that measures the degree of change in each token's representation. This seamlessly integrates token compression with FlashAttention, without attention maps or retraining. Our method further generalizes beyond Transformers to CNNs and state space models. Extensive experiments show that Representation Shift enables effective token compression compatible with FlashAttention, yielding significant speedups of up to 5.5% and 4.4% in video-text retrieval and video QA, respectively. Code is available at https://github.com/mlvlab/Representation-Shift.

Text-Video Retrieval aims to find the most relevant text (or video) candidate given a video (or text) query from large-scale online databases. Recent work leverages multi-modal large language models (MLLMs) to improve retrieval, especially for long or complex query-candidate pairs. However, we observe that the naive application of MLLMs, i.e., retrieval based on candidate likelihood, introduces candidate prior bias, favoring candidates with inherently higher priors over those more relevant to the query. To this end, we propose a novel retrieval framework, Bidirectional Likelihood Estimation with MLLM (BLiM), which leverages both query and candidate likelihoods by training the model to generate text from a given video as well as video features from a given text. Furthermore, we introduce Candidate Prior Normalization (CPN), a simple yet effective training-free score calibration module designed to mitigate candidate prior bias in candidate likelihood. On four Text-Video Retrieval benchmarks, our BLiM equipped with CPN outperforms previous state-of-the-art models by 6.4 R@1 on average, effectively alleviating candidate prior bias and emphasizing query-candidate relevance. Our in-depth analysis across various multi-modal tasks beyond retrieval highlights the broad applicability of CPN which enhances visual understanding by reducing reliance on textual priors. Code is available at https://github.com/mlvlab/BLiM.

Bayesian Optimization (BO) has been recognized for its effectiveness in optimizing expensive and complex objective functions. Recent advancements in Latent Bayesian Optimization (LBO) have shown promise by integrating generative models such as variational autoencoders (VAEs) to manage the complexity of high-dimensional and structured data spaces. However, existing LBO approaches often suffer from the value discrepancy problem, which arises from the reconstruction gap between input and latent spaces. This value discrepancy problem propagates errors throughout the optimization process, leading to suboptimal outcomes. To address this issue, we propose a Normalizing Flow-based Bayesian Optimization (NF-BO), which utilizes normalizing flow as a generative model to establish one-to-one encoding function from the input space to the latent space, along with its left-inverse decoding function, eliminating the reconstruction gap. Specifically, we introduce SeqFlow, an autoregressive normalizing flow for sequence data. In addition, we develop a new candidate sampling strategy that dynamically adjusts the exploration probability for each token based on its importance. Through extensive experiments, our NF-BO method demonstrates superior performance in molecule generation tasks, significantly outperforming both traditional and recent LBO approaches.

Open-source pre-trained models hold great potential for diverse applications, but their utility declines when their training data is unavailable. Data-Free Image Synthesis (DFIS) aims to generate images that approximate the learned data distribution of a pre-trained model without accessing the original data. However, existing DFIS methods produce samples that deviate from the training data distribution due to the lack of prior knowledge about natural images. To overcome this limitation, we propose DDIS, the first Diffusion-assisted Data-free Image Synthesis method that leverages a text-to-image diffusion model as a powerful image prior, improving synthetic image quality. DDIS extracts knowledge about the learned distribution from the given model and uses it to guide the diffusion model, enabling the generation of images that accurately align with the training data distribution. To achieve this, we introduce Domain Alignment Guidance (DAG) that aligns the synthetic data domain with the training data domain during the diffusion sampling process. Furthermore, we optimize a single Class Alignment Token (CAT) embedding to effectively capture class-specific attributes in the training dataset. Experiments on PACS and ImageNet demonstrate that DDIS outperforms prior DFIS methods by generating samples that better reflect the training data distribution, achieving SOTA performance in data-free applications.

Flow and diffusion models have demonstrated strong performance and training stability across various tasks but lack two critical properties of simulation-based methods: freedom of dimensionality and adaptability to different inference trajectories. To address this limitation, we propose the Multidimensional Adaptive Coefficient (MAC), a plug-in module for flow and diffusion models that extends conventional unidimensional coefficients to multidimensional ones and enables inference trajectory-wise adaptation. MAC is trained via simulation-based feedback through adversarial refinement. Empirical results across diverse frameworks and datasets demonstrate that MAC enhances generative quality with high training efficiency. Consequently, our work offers a new perspective on inference trajectory optimality, encouraging future research to move beyond vector field design and to leverage training-efficient, simulation-based optimization.

Multimodal Large Language Models (MLLMs) have shown promising progress in understanding and analyzing video content. However, processing long videos remains a significant challenge constrained by LLM's context size. To address this limitation, we propose \textbf{LongVU}, a spatiotemporal adaptive compression mechanism that reduces the number of video tokens while preserving visual details of long videos. Our idea is based on leveraging cross-modal query and inter-frame dependencies to adaptively reduce temporal and spatial redundancy in videos. Specifically, we leverage DINOv2 features to remove redundant frames that exhibit high similarity. Then we utilize text-guided cross-modal query for selective frame feature reduction. Further, we perform spatial token reduction across frames based on their temporal dependencies. Our adaptive compression strategy effectively processes a large number of frames with little visual information loss within given context length. Our LongVU consistently surpass existing methods across a variety of video understanding benchmarks, especially on hour-long video understanding tasks such as VideoMME and MLVU. Given a light-weight LLM, our LongVU also scales effectively into a smaller size with state-of-the-art video understanding performance.

Large language models (LLMs) have achieved remarkable success across diverse domains, due to their strong instruction-following capabilities. This raised interest in optimizing instructions for black-box LLMs, whose internal parameters are inaccessible but popular for their strong performance and ease of use. Recent approaches leverage white-box LLMs to assist instruction optimization for black-box LLMs by generating instructions from soft prompts. However, white-box LLMs often map different soft prompts to the same instruction, leading to redundant queries to the black-box model. While previous studies regarded this many-to-one mapping as a redundancy to be avoided, we reinterpret it as useful prior knowledge that can enhance the optimization performance. To this end, we introduce PREimage-informed inSTruction Optimization (PRESTO), a novel framework that leverages the preimage structure of soft prompts to improve query efficiency. PRESTO consists of three key components: (1) score sharing, which shares the evaluation score with all soft prompts in a preimage; (2) preimage-based initialization, which select initial data points that maximize search space coverage using preimage information; and (3) score consistency regularization, which enforces prediction consistency within each preimage. By leveraging preimages, PRESTO observes 14 times more scored data under the same query budget, resulting in more efficient optimization. Experimental results on 33 instruction optimization tasks demonstrate the superior performance of PRESTO.

Recently, Flow Matching models have pushed the boundaries of high-fidelity data generation across a wide range of domains. It typically employs a single large network to learn the entire generative trajectory from noise to data. Despite their effectiveness, this design struggles to capture distinct signal characteristics across timesteps simultaneously and incurs substantial inference costs due to the iterative evaluation of the entire model. To address these limitations, we propose Blockwise Flow Matching (BFM), a novel framework that partitions the generative trajectory into multiple temporal segments, each modeled by smaller but specialized velocity blocks. This blockwise design enables each block to specialize effectively in its designated interval, improving inference efficiency and sample quality. To further enhance generation fidelity, we introduce a Semantic Feature Guidance module that explicitly conditions velocity blocks on semantically rich features aligned with pretrained representations. Additionally, we propose a lightweight Feature Residual Approximation strategy that preserves semantic quality while significantly reducing inference cost. Extensive experiments on ImageNet 256x256 demonstrate that BFM establishes a substantially improved Pareto frontier over existing Flow Matching methods, achieving 2.1x to 4.9x accelerations in inference complexity at comparable generation performance.

Recent works have demonstrated the effectiveness of reinforcement learning (RL)-based post-training for enhancing the reasoning capabilities of large language models (LLMs). In particular, Group Relative Policy Optimization (GRPO) has shown impressive success using a PPO-style reinforcement learning algorithm with group-normalized rewards. However, the effectiveness of GRPO in Video Large Language Models (VideoLLMs) remains underexplored. In this paper, we explore GRPO and identify two issues that hinder effective learning: (1) reliance on safeguards, and (2) vanishing advantage. To mitigate these challenges, we propose DeepVideo-R1, a video large language model trained with Reg-GRPO (Regressive GRPO) and difficulty-aware data augmentation. Reg-GRPO reformulates the GRPO loss function as a regression task that directly predicts the advantage in GRPO, eliminating the need for safeguards such as clipping and min operations. This directly aligns the model with the advantages, providing guidance to prefer better outputs. The difficulty-aware data augmentation strategy augments input prompts/videos to target solvable difficulty levels, enabling diverse reward signals. Our experimental results show that our approach significantly improves video reasoning performance across multiple benchmarks.

Zero-shot Human-Object Interaction detection aims to localize humans and objects in an image and recognize their interaction, even when specific verb-object pairs are unseen during training. Recent works have shown promising results using prompt learning with pretrained vision-language models such as CLIP, which align natural language prompts with visual features in a shared embedding space. However, existing approaches still fail to handle the *visual complexity of interaction*—including (1) *intra-class visual diversity*, where instances of the same verb appear in diverse poses and contexts, and (2) *inter-class visual entanglement*, where distinct verbs yield visually similar patterns. To address these challenges, we propose **VDRP**, a framework for *Visual Diversity and Region-aware Prompt learning*. First, we introduce a visual diversity-aware prompt learning strategy that injects group-wise visual variance into the context embedding. We further apply Gaussian perturbation to encourage the prompt to capture diverse visual variations of a verb. Second, we retrieve region-specific concepts from the human, object, and union regions. These are used to augment the diversity-aware prompt embeddings, yielding region-aware prompts that improve verb-level discrimination. Experiments on the HICO-DET benchmark demonstrate that our method achieves state-of-the-art performance under four zero-shot evaluation settings, effectively addressing both intra-class diversity and inter-class visual entanglement. Code is available at https://github.com/mlvlab/VDRP.

Transformer-based models have recently shown success in representation learning on graph-structured data beyond natural language processing and computer vision. However, the success is limited to small-scale graphs due to the drawbacks of full dot-product attention on graphs such as the quadratic complexity with respect to the number of nodes and message aggregation from enormous irrelevant nodes. To address these issues, we propose Deformable Graph Transformer (DGT) that performs sparse attention via dynamically selected relevant nodes for efficiently handling large-scale graphs with a linear complexity in the number of nodes. Specifically, our framework first constructs multiple node sequences with various criteria to consider both structural and semantic proximity. Then, combining with our learnable Katz Positional Encodings, the sparse attention is applied to the node sequences for learning node representations with a significantly reduced computational cost. Extensive experiments demonstrate that our DGT achieves superior performance on 7 graph benchmark datasets with 2.5 $\sim$∼ 449 times less computational cost compared to transformer-based graph models with full attention.

Video Transformers have become the prevalent solution for various video downstream tasks with superior expressive power and flexibility. However, these video transformers suffer from heavy computational costs induced by the massive number of tokens across the entire video frames, which has been the major barrier to train and deploy the model. Further, the patches irrelevant to the main contents, e.g., backgrounds, degrade the generalization performance of models. To tackle these issues, we propose training-free token merging for lightweight video Transformer (vid-TLDR) that aims to enhance the efficiency of video Transformers by merging the background tokens without additional training. For vid-TLDR, we introduce a novel approach to capture the salient regions in videos only with the attention map. Further, we introduce the saliency-aware token merging strategy by dropping the background tokens and sharpening the object scores. Our experiments show that vid-TLDR significantly mitigates the computational complexity of video Transformers while achieving competitive performance compared to the base model without vid-TLDR. Code is available at https://github.com/mlvlab/vid-TLDR.

Open-vocabulary object detection (OVD) has been stud-ied with Vision-Language Models (VLMs) to detect novel objects beyond the pre-trained categories. Previous ap-proaches improve the generalization ability to expand the knowledge of the detector, using ‘positive’ pseudo-labels with additional ‘class' names, e.g., sock, iPod, and alli-gator. To extend the previous methods in two aspects, we propose Retrieval-Augmented Losses and visual Features (RALF). Our method retrieves related ‘negative’ classes and augments loss functions. Also, visual features are aug-mented with ‘verbalized concepts' of classes, e.g., worn on the feet, handheld music player, and sharp teeth. Specif-ically, RALF consists of two modules: Retrieval Aug-mented Losses (RAL) and Retrieval-Augmented visual Fea-tures (RAF). RAL constitutes two losses reflecting the se-mantic similarity with negative vocabularies. In addition, RAF augments visual features with the verbalized con-cepts from a large language model (LLM). Our experiments demonstrate the effectiveness of RALF on COCO and LVIS benchmark datasets. We achieve improvement up to 3.4 box APN50on novel categories of the COCO dataset and 3.6 mask APr gains on the LVIS dataset. Code is available at https://github.com/mlvlab/RALF.

Visual Relationship Detection (VRD) has seen significant advancements with Transformer-based architectures recently. However, we identify two key limitations in a conventional label assignment for training Transformer-based VRD models, which is a process of mapping a ground-truth (GT) to a prediction. Under the conventional assignment, an ‘unspecialized’ query is trained since a query is expected to detect every relation, which makes it difficult for a query to specialize in specific relations. Furthermore, a query is also insufficiently trained since a GT is assigned only to a single prediction, therefore near-correct or even correct predictions are suppressed by being assigned ‘no relation (⊘)‘ as a GT. To address these issues, we propose Groupwise Query Specialization and Quality-Aware Multi-Assignment (SpeaQ). Groupwise Query Specialization trains a ‘specialized’ query by dividing queries and relations into disjoint groups and directing a query in a specific query group solely toward relations in the corresponding relation group. Quality-Aware Multi-Assignment further facilitates the training by assigning a GT to multiple predictions that are significantly close to a GT in terms of a subject, an object, and the relation in between. Experimental results and analyses show that SpeaQ effectively trains ‘specialized’ queries, which better utilize the capacity of a model, resulting in consistent performance gains with ‘zero’ additional inference cost across multiple VRD models and benchmarks. Code is available at https://github.com/m1vlab/SpeaQ.

Vision Transformer (ViT) has emerged as a prominent backbone for computer vision. For more efficient ViTs, re-cent works lessen the quadratic cost of the self-attention layer by pruning or fusing the redundant tokens. How-ever, these works faced the speed-accuracy trade-off caused by the loss of information. Here, we argue that token fusion needs to consider diverse relations between tokens to minimize information loss. In this paper, we propose a Multi-criteria Token Fusion (MCTF), that gradually fuses the tokens based on multi-criteria (i.e., similarity, informativeness, and size of fused tokens). Further, we utilize the one-step-ahead attention, which is the improved approach to capture the informativeness of the tokens. By training the model equipped with MCTF using a token reduction consistency, we achieve the best speed-accuracy trade-off in the image classification (ImageNet1K). Experimental re-sults prove that MCTF consistently surpasses the previous reduction methods with and without training. Specifically, DeiT-T and DeiT-S with MCTF reduce FLOPs by about 44% while improving the performance (+0.5%, and +0.3%) over the base model, respectively. We also demonstrate the applicability of MCTF in various Vision Transformers (e.g., T2T-ViT, LV-ViT), achieving at least 31% speedup without performance degradation. Code is available at https://github.com/mlvlab/MCTF.

Pre-trained vision-language models have shown impres-sive success on various computer vision tasks with their zero-shot generalizability. Recently, prompt learning approaches have been explored to efficiently and effectively adapt the vision-language models to a variety of down-stream tasks. However, most existing prompt learning meth-ods suffer from task overfitting since the general knowl-edge of the pre-trained vision language models is forgot-ten while the prompts are finetuned on a small data set from a specific target task. To address this issue, we pro-pose a Prompt Meta-Regularization (ProMetaR) to im-prove the generalizability of prompt learning for vision-language models. Specifically, ProMetaR meta-learns both the regularizer and the soft prompts to harness the task-specific knowledge from the downstream tasks and task-agnostic general knowledge from the vision-language mod-els. Further, ProMetaR augments the task to gener-ate multiple virtual tasks to alleviate the meta-overfitting. In addition, we provide the analysis to comprehend how ProMetaR improves the generalizability of prompt tuning in the perspective of the gradient alignment. Our exten-sive experiments demonstrate that our ProMetaR improves the generalizability of conventional prompt learning meth-ods under base-to-base/base-to-new and domain general-ization settings. The code of ProMetaR is available at https://github.com/mlvlab/ProMetaR.

Knowledge graph-grounded dialog generation requires retrieving a dialog-relevant subgraph from the given knowledge base graph and integrating it with the dialog history.Previous works typically represent the graph using an external encoder, such as graph neural networks, and retrieve relevant triplets based on the similarity between single-vector representations of triplets and the dialog history.However, these external encoders fail to leverage the rich knowledge of pretrained language models, and the retrieval process is also suboptimal due to the information bottleneck caused by the single-vector abstraction of the dialog history.In this work, we propose Dialog generation with Generative Subgraph Retrieval (DialogGSR), which retrieves relevant knowledge subgraphs by directly generating their token sequences on top of language models.For effective generative subgraph retrieval, we introduce two key methods: (i) structureaware knowledge graph linearization with selfsupervised graph-specific tokens and (ii) graphconstrained decoding utilizing graph structural proximity-based entity informativeness scores for valid and relevant generative retrieval.Di-alogGSR achieves state-of-the-art performance in knowledge graph-grounded dialog generation, as demonstrated on OpenDialKG and KOMODIS datasets.

Recent studies have introduced a new class of generative models for synthesizing implicit neural representations (INRs) that capture arbitrary continuous signals in various domains. These models opened the door for domain-agnostic generative models, but they often fail to achieve high-quality generation. We observed that the existing methods generate the weights of neural networks to parameterize INRs and evaluate the network with fixed positional embeddings (PEs). Arguably, this architecture limits the expressive power of generative models and results in low-quality INR generation. To address this limitation, we propose Domain-agnostic Latent Diffusion Model for INRs (DDMI) that generates adaptive positional embeddings instead of neural networks' weights. Specifically, we develop a Discrete-to-continuous space Variational AutoEncoder (D2C-VAE) that seamlessly connects discrete data and continuous signal functions in the shared latent space. Additionally, we introduce a novel conditioning mechanism for evaluating INRs with the hierarchically decomposed PEs to further enhance expressive power. Extensive experiments across four modalities, \eg, 2D images, 3D shapes, Neural Radiance Fields, and videos, with seven benchmark datasets, demonstrate the versatility of DDMI and its superior performance compared to the existing INR generative models. Code is available at \href{https://github.com/mlvlab/DDMI}{https://github.com/mlvlab/DDMI}.

Semantic image synthesis (SIS) is a task to generate realistic images corresponding to semantic maps (labels). However, in real-world applications, SIS often encounters noisy user inputs. To address this, we propose Stochastic Conditional Diffusion Model (SCDM), which is a robust conditional diffusion model that features novel forward and generation processes tailored for SIS with noisy labels. It enhances robustness by stochastically perturbing the semantic label maps through Label Diffusion, which diffuses the labels with discrete diffusion. Through the diffusion of labels, the noisy and clean semantic maps become similar as the timestep increases, eventually becoming identical at $t=T$. This facilitates the generation of an image close to a clean image, enabling robust generation. Furthermore, we propose a class-wise noise schedule to differentially diffuse the labels depending on the class. We demonstrate that the proposed method generates high-quality samples through extensive experiments and analyses on benchmark datasets, including a novel experimental setup simulating human errors during real-world applications. Code is available at https://github.com/mlvlab/SCDM.

Latent Bayesian optimization (LBO) approaches have successfully adopted Bayesian optimization over a continuous latent space by employing an encoder-decoder architecture to address the challenge of optimization in a high dimensional or discrete input space. LBO learns a surrogate model to approximate the black-box objective function in the latent space. However, we observed that most LBO methods suffer from the `misalignment problem', which is induced by the reconstruction error of the encoder-decoder architecture. It hinders learning an accurate surrogate model and generating high-quality solutions. In addition, several trust region-based LBO methods select the anchor, the center of the trust region, based solely on the objective function value without considering the trust region's potential to enhance the optimization process. To address these issues, we propose $\textbf{Inv}$ersion-based Latent $\textbf{B}$ayesian $\textbf{O}$ptimization (InvBO), a plug-and-play module for LBO. InvBO consists of two components: an inversion method and a potential-aware trust region anchor selection. The inversion method searches the latent code that completely reconstructs the given target data. The potential-aware trust region anchor selection considers the potential capability of the trust region for better local optimization. Experimental results demonstrate the effectiveness of InvBO on nine real-world benchmarks, such as molecule design and arithmetic expression fitting tasks. Code is available at https://github.com/mlvlab/InvBO.

Large Language Models (LLMs) have demonstrated remarkable generalization and instruction-following capabilities with instruction tuning. The advancements in LLMs and instruction tuning have led to the development of Large Vision-Language Models (LVLMs). However, the competency of the LLMs and instruction tuning have been less explored in the molecular domain. Thus, we propose LLaMo: Large Language Model-based Molecular graph assistant, which is an end-to- end trained large molecular graph-language model. To bridge the discrepancy between the language and graph modalities, we present the multi-level graph projector that transforms graph representations into graph tokens by abstracting the output representations of each GNN layer and motif representations with the cross-attention mechanism. We also introduce machine-generated molecular graph instruction data to instruction-tune the large molecular graph-language model for general-purpose molecule and language understanding. Our extensive experiments demonstrate that LLaMo shows the best performance on diverse tasks, such as molecular description generation, property prediction, and IUPAC name prediction. The code of LLaMo is available at https://github.com/mlvlab/LLaMo.

Rectified flow and reflow procedures have significantly advanced fast generation by progressively straightening ordinary differential equation (ODE) flows under the assumption that image and noise pairs, known as coupling, can be approximated by straight trajectories with constant velocity. However, we observe that the constant velocity modeling and reflow procedures have limitations in accurately learning to couple with flow crossing, leading to suboptimal few-step generation. To overcome the limitations, we introduce the Constant Acceleration Flow (CAF), a novel framework based on a simple constant acceleration equation. Additionally, we propose two techniques to improve estimation accuracy: initial velocity conditioning for the acceleration model and a reflow process for the initial velocity. Our comparative studies show that CAF not only outperforms rectified flow with reflow procedures in terms of speed and accuracy but also demonstrates substantial improvements in preserving coupling for fast generation.

A cache side-channel attack retrieves victim’s sensitive information from a system by exploiting shared cache of CPUs. Since conventional cache side-channel attacks such as FLUSH+RELOAD and PRIME+PROBE are likely to incur numerous cache events, such as cache hits and misses, many previous strategies have focused on monitoring cache events for attack detection. However, as recently proposed attacks such as PRIME+ABORT have exploited the other events as side-channels, it has become challenging to detect them by monitoring only cache events. In this paper, we investigate PRIME+ABORT attack and identifies Intel TSX hardware events are tightly coupled with it as well as cache events. Based on our finding, we propose a novel deep learning-based cache side-channel attack detection method called FRIME. It can concurrently detect not only the conventional attacks such as FLUSH+RELOAD, PRIME+PROBE, but also PRIME+ABORT by leveraging both event types. In order to demonstrate the efficacy of our cache side-channel attack detection scheme in diverse workload conditions in the real world, we implement it using MLP, RNN, and LSTM deep learning models, demonstrating LSTM-based method outperforms the other implementations in terms of detection accuracy.

Question Answering (QA) is a task that entails reasoning over natural language contexts, and many relevant works augment language models (LMs) with graph neural networks (GNNs) to encode the Knowledge Graph (KG) information. However, most existing GNN-based modules for QA do not take advantage of rich relational information of KGs and depend on limited information interaction between the LM and the KG. To address these issues, we propose Question Answering Transformer (QAT), which is designed to jointly reason over language and graphs with respect to entity relations in a unified manner. Specifically, QAT constructs Meta-Path tokens, which learn relation-centric embeddings based on diverse structural and semantic relations. Then, our Relation-Aware Self-Attention module comprehensively integrates different modalities via the Cross-Modal Relative Position Bias, which guides information exchange between relevant entities of different modalities. We validate the effectiveness of QAT on commonsense question answering datasets like CommonsenseQA and OpenBookQA, and on a medical question answering dataset, MedQA-USMLE. On all the datasets, our method achieves state-of-the-art performance. Our code is available at http://github.com/mlvlab/QAT.

Foundation models have shown outstanding performance and generalization capabilities across domains. Since most studies on foundation models mainly focus on the pretraining phase, a naive strategy to minimize a single task-specific loss is adopted for fine-tuning. However, such fine-tuning methods do not fully leverage other losses that are potentially beneficial for the target task. Therefore, we propose MEta Loss TRansformer (MELTR), a plug-in module that automatically and non-linearly combines various loss functions to aid learning the target task via auxiliary learning. We formulate the auxiliary learning as a bi-level optimization problem and present an efficient optimization algorithm based on Approximate Implicit Differentiation (AID). For evaluation, we apply our framework to various video foundation models (UniVL, Violet and All-in-one), and show significant performance gain on all four downstream tasks: text-to-video retrieval, video question answering, video captioning, and multimodal sentiment analysis. Our qualitative analyses demonstrate that MELTR adequately 'transforms' individual loss functions and 'melts' them into an effective unified loss. Code is available at https://github.com/mlvlab/MELTR.

Transformers have shown superior performance on various computer vision tasks with their capabilities to capture long-range dependencies. Despite the success, it is challenging to directly apply Transformers on point clouds due to their quadratic cost in the number of points. In this paper, we present a Self-Positioning point-based Transformer (SPoTr), which is designed to capture both local and global shape contexts with reduced complexity. Specifically, this architecture consists of local self-attention and self-positioning point-based global cross-attention. The self-positioning points, adaptively located based on the input shape, consider both spatial and semantic information with disentangled attention to improve expressive power. With the self-positioning points, we propose a novel global cross-attention mechanism for point clouds, which improves the scalability of global self-attention by allowing the attention module to compute attention weights with only a small set of self-positioning points. Experiments show the effectiveness of SPoTr on three point cloud tasks such as shape classification, part segmentation, and scene segmentation. In particular, our proposed model achieves an accuracy gain of 2.6% over the previous best models on shape classification with ScanObjectNN. We also provide qualitative analyses to demonstrate the interpretability of self-positioning points. The code of SPoTr is available at https://github.com/mlvlab/SPoTr.

Pre-trained vision-language models (VLMs) have shown impressive performance on various downstream tasks by utilizing knowledge learned from large data. In general, the performance of VLMs on target tasks can be further improved by prompt tuning, which adds context to the input image or text. By leveraging data from target tasks, various prompt-tuning methods have been studied in the literature. A key to prompt tuning is the feature space alignment between two modalities via learnable vectors with model parameters fixed. We observed that the alignment becomes more effective when embeddings of each modality are ‘well-arranged’ in the latent space. Inspired by this observation, we proposed distribution-aware prompt tuning (DAPT) for vision-language models, which is simple yet effective. Specifically, the prompts are learned by maximizing inter-dispersion, the distance between classes, as well as minimizing the intra-dispersion measured by the distance between embeddings from the same class. Our extensive experiments on 11 benchmark datasets demonstrate that our method significantly improves generalizability. The code is available at https://github.com/mlvlab/DAPT.

Learning implicit templates as neural fields has recently shown impressive performance in unsupervised shape correspondence. Despite the success, we observe current approaches, which solely rely on geometric information, often learn suboptimal deformation across generic object shapes, which have high structural variability. In this paper, we highlight the importance of part deformation consistency and propose a semantic-aware implicit template learning framework to enable semantically plausible deformation. By leveraging semantic prior from a self-supervised feature extractor, we suggest local conditioning with novel semantic-aware deformation code and deformation consistency regularizations regarding part deformation, global deformation, and global scaling. Our extensive experiments demonstrate the superiority of the proposed method over baselines in various tasks: keypoint transfer, part label transfer, and texture transfer. More interestingly, our framework shows a larger performance gain under more challenging settings. We also provide qualitative analyses to validate the effectiveness of semantic-aware deformation. The code is available at https://github.com/mlvlab/PDC.

Video Question Answering (VideoQA) is a challenging task that entails complex multi-modal reasoning. In contrast to multiple-choice VideoQA which aims to predict the answer given several options, the goal of open-ended VideoQA is to answer questions without restricting candidate answers. However, the majority of previous VideoQA models formulate open-ended VideoQA as a classification task to classify the video-question pairs into a fixed answer set, i.e., closed-vocabulary, which contains only frequent answers (e.g., top-1000 answers). This leads the model to be biased toward only frequent answers and fail to generalize on out-of-vocabulary answers. We hence propose a new benchmark, Open-vocabulary Video Question Answering (OVQA), to measure the generalizability of VideoQA models by considering rare and unseen answers. In addition, in order to improve the model’s generalization power, we introduce a novel GNN-based soft verbalizer that enhances the prediction on rare and unseen answers by aggregating the information from their similar words. For evaluation, we introduce new baselines by modifying the existing (closed-vocabulary) open-ended VideoQA models and improve their performances by further taking into account rare and unseen answers. Our ablation studies and qualitative analyses demonstrate that our GNN-based soft verbalizer further improves the model performance, especially on rare and unseen answers. We hope that our benchmark OVQA can serve as a guide for evaluating the generalizability of VideoQA models and inspire future research. Code is available at https://github.com/mlvlab/OVQA.

In recent years, prompt tuning has proven effective in adapting pre-trained vision-language models to downstream tasks. These methods aim to adapt the pre-trained models by introducing learnable prompts while keeping pretrained weights frozen. However, learnable prompts can affect the internal representation within the self-attention module, which may negatively impact performance variance and generalization, especially in data-deficient settings. To address these issues, we propose a novel approach, Read-only Prompt Optimization (RPO). RPO leverages masked attention to prevent the internal representation shift in the pre-trained model. Further, to facilitate the optimization of RPO, the read-only prompts are initialized based on special tokens of the pre-trained model. Our extensive experiments demonstrate that RPO outperforms CLIP and CoCoOp in base-to-new generalization and domain generalization while displaying better robustness. Also, the proposed method achieves better generalization on extremely data-deficient settings, while improving parameter efficiency and computational overhead. Code is available at https://github.com/mlvlab/RPO.

Multi-resolution hash encoding has recently been proposed to reduce the computational cost of neural renderings, such as NeRF. This method requires accurate camera poses for the neural renderings of given scenes. However, contrary to previous methods jointly optimizing camera poses and 3D scenes, the naive gradient-based camera pose refinement method using multi-resolution hash encoding severely deteriorates performance. We propose a joint optimization algorithm to calibrate the camera pose and learn a geometric representation using efficient multi-resolution hash encoding. Showing that the oscillating gradient flows of hash encoding interfere with the registration of camera poses, our method addresses the issue by utilizing smooth interpolation weighting to stabilize the gradient oscillation for the ray samplings across hash grids. Moreover, the curriculum training procedure helps to learn the level-wise hash encoding, further increasing the pose refinement. Experiments on the novel-view synthesis datasets validate that our learning frameworks achieve state-of-the-art performance and rapid convergence of neural rendering.

Multi-hop Knowledge Graph Question Answering (KGQA) is a task that involves retrieving nodes from a knowledge graph (KG) to answer natural language questions. Recent GNN-based approaches formulate this task as a KG path searching problem, where messages are sequentially propagated from the seed node towards the answer nodes. However, these messages are past-oriented, and they do not consider the full KG context. To make matters worse, KG nodes often represent pronoun entities and are sometimes encrypted, being uninformative in selecting between paths. To address these problems, we propose Neural Tree Search (NuTrea), a tree search-based GNN model that incorporates the broader KG context. Our model adopts a message-passing scheme that probes the unreached subtree regions to boost the past-oriented embeddings. In addition, we introduce the Relation Frequency-Inverse Entity Frequency (RF-IEF) node embedding that considers the global KG context to better characterize ambiguous KG nodes. The general effectiveness of our approach is demonstrated through experiments on three major multi-hop KGQA benchmark datasets, and our extensive analyses further validate its expressiveness and robustness. Overall, NuTrea provides a powerful means to query the KG with complex natural language questions. Code is available at https://github.com/mlvlab/NuTrea.

Neural Radiance Field (NeRF) has enabled novel view synthesis with high fidelity given images and camera poses. Subsequent works even succeeded in eliminating the necessity of pose priors by jointly optimizing NeRF and camera pose. However, these works are limited to relatively simple settings such as photometrically consistent and occluder-free image collections or a sequence of images from a video. So they have difficulty handling unconstrained images with varying illumination and transient occluders. In this paper, we propose **UP-NeRF** (**U**nconstrained **P**ose-prior-free **Ne**ural **R**adiance **F**ields) to optimize NeRF with unconstrained image collections without camera pose prior. We tackle these challenges with surrogate tasks that optimize color-insensitive feature fields and a separate module for transient occluders to block their influence on pose estimation. In addition, we introduce a candidate head to enable more robust pose estimation and transient-aware depth supervision to minimize the effect of incorrect prior. Our experiments verify the superior performance of our method compared to the baselines including BARF and its variants in a challenging internet photo collection, *Phototourism dataset*. The code of UP-NeRF is available at https://github.com/mlvlab/UP-NeRF.

Bayesian optimization is a powerful method for optimizing black-box functions with limited function evaluations. Recent works have shown that optimization in a latent space through deep generative models such as variational autoencoders leads to effective and efficient Bayesian optimization for structured or discrete data. However, as the optimization does not take place in the input space, it leads to an inherent gap that results in potentially suboptimal solutions. To alleviate the discrepancy, we propose Correlated latent space Bayesian Optimization (CoBO), which focuses on learning correlated latent spaces characterized by a strong correlation between the distances in the latent space and the distances within the objective function. Specifically, our method introduces Lipschitz regularization, loss weighting, and trust region recoordination to minimize the inherent gap around the promising areas. We demonstrate the effectiveness of our approach on several optimization tasks in discrete data, such as molecule design and arithmetic expression fitting, and achieve high performance within a small budget.

Deep neural networks easily suffer from weak generalization caused by overfitting on biased data. One popular remedy to alleviate this issue is sample reweighting methods that adaptively adjust the importance of biased samples. Separate from the effort to reduce bias, recent works show that the generalization power can be improved by auxiliary tasks. Inspired by the two lines of works, we extend the sample reweighting methods to auxiliary tasks. In this paper, we propose a novel auxiliary learning framework that improves the primary task by adaptively adjusting the weights of samples from multiple tasks rather than samples from a single task using a weighting function. The weighting function is optimized by meta-learning along the gradient of the loss for meta-data, which is a small unbiased validation data. We also present a task-activation score that indicates the correlation between the learning tendency of the training samples and meta-data samples. This score is utilized as a regularizer for meta-learning objective. Our framework can obtain powerful representations for the primary task on biased data by automatically identifying effective combinations of tasks. Our experiments demonstrate that our proposed method consistently outperforms all baselines and state-of-the-art methods on both corrupted labels and class imbalance settings.

Many self-supervised representation learning methods have achieved high performance in image classification tasks. However, these methods have limited performance on localization tasks such as object detection or semantic segmentation. Most self-supervised representation learning methods are optimized with only one global representation, which does not pay much attention to the spatial information in an image. We propose a simple and effective method that uses the positional relationships between the entities in an image by shuffling the convolution kernels. Our method extends current self-supervised learning and calculates the pixel-wise (dis) similarities between the output of the standard convolution kernels and that of the randomly shuffled convolution kernels. Our proposed method achieves higher performance on object detection, instance segmentation, and semantic segmentation when attached to recent self-supervised learning methods.

Graph neural networks (GNNs) have significantly improved the representation power for graph-structured data. Despite of the recent success of GNNs, the graph convolution in most GNNs have two limitations. Since the graph convolution is performed in a small local neighborhood on the input graph, it is inherently incapable to capture long-range dependencies between distance nodes. In addition, when a node has neighbors that belong to different classes, i.e., heterophily, the aggregated messages from them often negatively affect representation learning. To address the two common problems of graph convolution, in this paper, we propose Deformable Graph Convolutional Networks (Deformable GCNs) that adaptively perform convolution in multiple latent spaces and capture short/long-range dependencies between nodes. Separated from node representations (features), our framework simultaneously learns the node positional embeddings (coordinates) to determine the relations between nodes in an end-to-end fashion. Depending on node position, the convolution kernels are deformed by deformation vectors and apply different transformations to its neighbor nodes. Our extensive experiments demonstrate that Deformable GCNs flexibly handles the heterophily and achieve the best performance in node classification tasks on six heterophilic graph datasets. Our code is publicly available at https://github.com/mlvlab/DeformableGCN.

Learning generic joint representations for video and text by a supervised method requires a prohibitively substantial amount of manually annotated video datasets. As a practical alternative, a large-scale but uncurated and narrated video dataset, HowTo100M, has recently been introduced. But it is still challenging to learn joint embeddings of video and text in a self-supervised manner, due to its ambiguity and non-sequential alignment. In this paper, we propose a novel multi-modal self-supervised framework Video-Text Temporally Weak Alignment-based Contrastive Learning (VT-TWINS) to capture significant information from noisy and weakly correlated data using a variant of Dynamic Time Warping (DTW). We observe that the standard DTW inherently cannot handle weakly correlated data and only considers the globally optimal alignment path. To address these problems, we develop a differentiable DTW which also reflects local information with weak temporal alignment. Moreover, our proposed model applies a contrastive learning scheme to learn feature representations on weakly correlated data. Our extensive experiments demonstrate that VT-TWINS attains significant improvements in multi-modal representation learning and outperforms various challenging downstream tasks. Code is available at https://github.com/mlvlab/VT-Twins.

Human-Object Interaction detection is a holistic visual recognition task that entails object detection as well as interaction classification. Previous works of HOI detection has been addressed by the various compositions of subset predictions, e.g., Image → HO → I, Image → HI →$O$. Recently, transformer based architecture for HOI has emerged, which directly predicts the HOI triplets in an end-to-end fashion (Image → HOI). Motivated by various inference paths for HOI detection, we propose cross-path consistency learning (CPC), which is a novel end-to-end learning strategy to improve HOI detection for transformers by leveraging augmented decoding paths. CPC learning enforces all the possible predictions from permuted inference sequences to be consistent. This simple scheme makes the model learn consistent representations, thereby improving generalization without increasing model capacity. Our experiments demonstrate the effectiveness of our method, and we achieved significant improvement on V-COCO and HICO-DET compared to the baseline models. Our code is available at https://github.com/mlvlab/CPChoi.

-SALSA leverages a new technique, called local style alignment, to generate a synthetic average that maximizes the retention of fine-grain visual patterns in the source images, thus improving the clinical utility of the generated images. On two benchmark datasets of diabetic retinopathy (EyePACS and APTOS), we demonstrate our improvement upon existing methods with respect to image fidelity, classification performance, and mitigation of membership inference attacks. Our work represents a step toward broader sharing of retinal images for scientific collaboration. Code is available at https://github.com/hcholab/k-salsa.

Normalizing flows model probability distributions by learning invertible transformations that transfer a simple distribution into complex distributions. Since the architecture of ResNet-based normalizing flows is more flexible than that of coupling-based models, ResNet-based normalizing flows have been widely studied in recent years. Despite their architectural flexibility, it is well-known that the current ResNet-based models suffer from constrained Lipschitz constants. In this paper, we propose the monotone formulation to overcome the issue of the Lipschitz constants using monotone operators and provide an in-depth theoretical analysis. Furthermore, we construct an activation function called Concatenated Pila (CPila) to improve gradient flow. The resulting model, Monotone Flows, exhibits an excellent performance on multiple density estimation benchmarks (MNIST, CIFAR-10, ImageNet32, ImageNet64). Code is available at https://github.com/mlvlab/MonotoneFlows.

Mixup is a commonly adopted data augmentation technique for image classification. Recent advances in mixup methods primarily focus on mixing based on saliency. However, many saliency detectors require intense computation and are especially burdensome for parameter-heavy transformer models. To this end, we propose TokenMixup, an efficient attention-guided token-level data augmentation method that aims to maximize the saliency of a mixed set of tokens. TokenMixup provides ×15 faster saliency-aware data augmentation compared to gradient-based methods. Moreover, we introduce a variant of TokenMixup which mixes tokens within a single instance, thereby enabling multi-scale feature augmentation. Experiments show that our methods significantly improve the baseline models’ performance on CIFAR and ImageNet-1K, while being more efficient than previous methods. We also reach state-of-the-art performance on CIFAR-100 among from-scratch transformer models. Code is available at https://github.com/mlvlab/TokenMixup.

This paper proposes Mutual Information Regularized Assignment (MIRA), a pseudo-labeling algorithm for unsupervised representation learning inspired by information maximization. We formulate online pseudo-labeling as an optimization problem to find pseudo-labels that maximize the mutual information between the label and data while being close to a given model probability. We derive a fixed-point iteration method and prove its convergence to the optimal solution. In contrast to baselines, MIRA combined with pseudo-label prediction enables a simple yet effective clustering-based representation learning without incorporating extra training techniques or artificial constraints such as sampling strategy, equipartition constraints, etc. With relatively small training epochs, representation learned by MIRA achieves state-of-the-art performance on various downstream tasks, including the linear/${\it k}$-NN evaluation and transfer learning. Especially, with only 400 epochs, our method applied to ImageNet dataset with ResNet-50 architecture achieves 75.6% linear evaluation accuracy.

Data augmentation is key to improving the generalization ability of deep learning models. Mixup is a simple and widely-used data augmentation technique that has proven effective in alleviating the problems of overfitting and data scarcity. Also, recent studies of saliency-aware Mixup in the image domain show that preserving discriminative parts is beneficial to improving the generalization performance. However, these Mixup-based data augmentations are underexplored in 3D vision, especially in point clouds. In this paper, we propose SageMix, a saliency-guided Mixup for point clouds to preserve salient local structures. Specifically, we extract salient regions from two point clouds and smoothly combine them into one continuous shape. With a simple sequential sampling by re-weighted saliency scores, SageMix preserves the local structure of salient regions. Extensive experiments demonstrate that the proposed method consistently outperforms existing Mixup methods in various benchmark point cloud datasets. With PointNet++, our method achieves an accuracy gain of 2.6% and 4.0% over standard training in ModelNet40 and ScanObjectNN, respectively. In addition to generalization performance, SageMix improves robustness and uncertainty calibration. Moreover, when adopting our method to various tasks including part segmentation and standard image classification, our method achieves competitive performance. Code is available at https://github.com/mlvlab/SageMix.

BACKGROUND: DCLEAR is an R package used for single cell lineage reconstruction. The advances of CRISPR-based gene editing technologies have enabled the prediction of cell lineage trees based on observed edited barcodes from each cell. However, the performance of existing reconstruction methods of cell lineage trees was not accessed until recently. In response to this problem, the Allen Institute hosted the Cell Lineage Reconstruction Dream Challenge in 2020 to crowdsource relevant knowledge from across the world. Our team won sub-challenges 2 and 3 in the challenge competition. RESULTS: The DCLEAR package contained the R codes, which was submitted in response to sub-challenges 2 and 3. Our method consists of two steps: (1) distance matrix estimation and (2) the tree reconstruction from the distance matrix. We proposed two novel methods for distance matrix estimation as outlined in the DCLEAR package. Using our method, we find that two of the more sophisticated distance methods display a substantially improved level of performance compared to the traditional Hamming distance method. DCLEAR is open source and freely available from R CRAN and from under the GNU General Public License, version 3. CONCLUSIONS: DCLEAR is a powerful resource for single cell lineage reconstruction.

Graph Neural Networks (GNNs) have been widely applied to various fields due to their powerful representations of graph-structured data. Despite the success of GNNs, most existing GNNs are designed to learn node representations on the fixed and homogeneous graphs. The limitations especially become problematic when learning representations on a misspecified graph or a heterogeneous graph that consists of various types of nodes and edges. To address these limitations, we propose Graph Transformer Networks (GTNs) that are capable of generating new graph structures, which preclude noisy connections and include useful connections (e.g., meta-paths) for tasks, while learning effective node representations on the new graphs in an end-to-end fashion. We further propose enhanced version of GTNs, Fast Graph Transformer Networks (FastGTNs), that improve scalability of graph transformations. Compared to GTNs, FastGTNs are up to 230× and 150× faster in inference and training, and use up to 100× and 148× less memory while allowing the identical graph transformations as GTNs. In addition, we extend graph transformations to the semantic proximity of nodes allowing non-local operations beyond meta-paths. Extensive experiments on both homogeneous graphs and heterogeneous graphs show that GTNs and FastGTNs with non-local operations achieve the state-of-the-art performance for node classification tasks. The code is available: https://github.com/seongjunyun/Graph_Transformer_Networks.

Human-Object Interaction (HOI) detection is a task of identifying "a set of interactions" in an image, which involves the i) localization of the subject (i.e., humans) and target (i.e., objects) of interaction, and ii) the classification of the interaction labels. Most existing methods have indirectly addressed this task by detecting human and object instances and individually inferring every pair of the detected instances. In this paper, we present a novel framework, referred by HOTR, which directly predicts a set of 〈human, object, interaction〉 triplets from an image based on a transformer encoder-decoder architecture. Through the set prediction, our method effectively exploits the inherent semantic relationships in an image and does not require time-consuming post-processing which is the main bottleneck of existing methods. Our proposed algorithm achieves the state-of-the-art performance in two HOI detection benchmarks with an inference time under 1 ms after object detection.

Despite the extensive usage of point clouds in 3D vision, relatively limited data are available for training deep neural networks. Although data augmentation is a standard approach to compensate for the scarcity of data, it has been less explored in the point cloud literature. In this paper, we propose a simple and effective augmentation method called PointWOLF for point cloud augmentation. The proposed method produces smoothly varying non-rigid deformations by locally weighted transformations centered at multiple anchor points. The smooth deformations allow diverse and realistic augmentations. Furthermore, in order to minimize the manual efforts to search the optimal hyperparameters for augmentation, we present AugTune, which generates augmented samples of desired difficulties producing targeted confidence scores. Our experiments show our framework consistently improves the performance for both shape classification and part segmentation tasks. Particularly, with PointNet++, PointWOLF achieves the state-of-the-art 89.7 accuracy on shape classification with the real-world ScanObjectNN dataset. The code is available at https://github.com/mlvlab/PointWOLF.

Graph Neural Networks (GNNs) often suffer from weak-generalization due to sparsely labeled data despite their promising results on various graph-based tasks. Data augmentation is a prevalent remedy to improve the generalization ability of models in many domains. However, due to the non-Euclidean nature of data space and the dependencies between samples, designing effective augmentation on graphs is challenging. In this paper, we propose a novel framework Metropolis-Hastings Data Augmentation (MH-Aug) that draws augmented graphs from an explicit target distribution for semi-supervised learning. MH-Aug produces a sequence of augmented graphs from the target distribution enables flexible control of the strength and diversity of augmentation. Since the direct sampling from the complex target distribution is challenging, we adopt the Metropolis-Hastings algorithm to obtain the augmented samples. We also propose a simple and effective semi-supervised learning strategy with generated samples from MH-Aug. Our extensive experiments demonstrate that MH-Aug can generate a sequence of samples according to the target distribution to significantly improve the performance of GNNs.

Graph Neural Networks (GNNs) have been widely applied to various fields for learning over graph-structured data. They have shown significant improvements over traditional heuristic methods in various tasks such as node classification and graph classification. However, since GNNs heavily rely on smoothed node features rather than graph structure, they often show poor performance than simple heuristic methods in link prediction where the structural information, e.g., overlapped neighborhoods, degrees, and shortest paths, is crucial. To address this limitation, we propose Neighborhood Overlap-aware Graph Neural Networks (Neo-GNNs) that learn useful structural features from an adjacency matrix and estimate overlapped neighborhoods for link prediction. Our Neo-GNNs generalize neighborhood overlap-based heuristic methods and handle overlapped multi-hop neighborhoods. Our extensive experiments on Open Graph Benchmark datasets (OGB) demonstrate that Neo-GNNs consistently achieve state-of-the-art performance in link prediction.

Graph neural networks have shown superior performance in a wide range of applications providing a powerful representation of graph-structured data. Recent works show that the representation can be further improved by auxiliary tasks. However, the auxiliary tasks for heterogeneous graphs, which contain rich semantic information with various types of nodes and edges, have less explored in the literature. In this paper, to learn graph neural networks on heterogeneous graphs we propose a novel self-supervised auxiliary learning method using meta paths, which are composite relations of multiple edge types. Our proposed method is learning to learn a primary task by predicting meta-paths as auxiliary tasks. This can be viewed as a type of meta-learning. The proposed method can identify an effective combination of auxiliary tasks and automatically balance them to improve the primary task. Our methods can be applied to any graph neural networks in a plug-in manner without manual labeling or additional data. The experiments demonstrate that the proposed method consistently improves the performance of link prediction and node classification on heterogeneous graphs.

For a service robot to assist humans, it should interact with objects of varying sizes and shapes existing in an indoor environment. 3D object detection must be preceded to achieve this goal since it provides the robot with the ability to perceive visual information. Most of the existing methods are anchor-based and predict the bounding box close to the ground truth among multiple candidates. However, it is complex to compute Intersection over Union (IoU) and Non-Maximum Suppression (NMS) per each anchor box. Therefore, we propose keypoint-based monocular 3D object detection, where each object's center location is only needed for reproducing predicted 3D bounding box without extra computation of the anchor boxes. Our 3D object detection also works well even if images are rotated corresponding to the robot's head movement. To properly train our network, the object center is based on a projected 3D location instead of 2D to take advantage of the exact center position of the object. Furthermore, we apply data augmentation using a perspective transformation. The method facilitates adding a small perturbation to the camera rotation angle randomly. We use the SUN RGB-D dataset, which has images taken indoor scenes with camera rotations for training and test set. Our approach particularly shows that the errors of object center location based on a single image reduce 15.4% and 24.2%, respectively, compared to the method without data augmentation.

There is much interest in computer vision to utilize commodity hardware for gaze estimation. A number of papers have shown that algorithms based on deep convolutional architectures are approaching accuracies where streaming data from mass-market devices can offer good gaze tracking performance, although a gap still remains between what is possible and the performance users will expect in real deployments. We observe that one obvious avenue for improvement relates to a gap between some basic technical assumptions behind most existing approaches and the statistical properties of the data used for training. Specifically, most training datasets involve tens of users with a few hundreds (or more) repeated acquisitions per user. The non i.i.d. nature of this data suggests better estimation may be possible if the model explicitly made use of such “repeated measurements” from each user as is commonly done in classical statistical analysis using so-called mixed effects models. The goal of this paper is to adapt these “mixed effects” ideas from statistics within a deep neural network architecture for gaze estimation, based on eye images. Such a formulation seeks to specifically utilize information regarding the hierarchical structure of the training data - each node in the hierarchy is a user who provides tens or hundreds of repeated samples. This modification yields an architecture that offers state of the art performance on various publicly available datasets improving results by 10-20%.

Graph neural networks (GNNs) have been widely used in representation learning on graphs and achieved state-of-the-art performance in tasks such as node classification and link prediction. However, most existing GNNs are designed to learn node representations on the fixed and homogeneous graphs. The limitations especially become problematic when learning representations on a misspecified graph or a heterogeneous graph that consists of various types of nodes and edges. In this paper, we propose Graph Transformer Networks (GTNs) that are capable of generating new graph structures, which involve identifying useful connections between unconnected nodes on the original graph, while learning effective node representation on the new graphs in an end-to-end fashion. Graph Transformer layer, a core layer of GTNs, learns a soft selection of edge types and composite relations for generating useful multi-hop connections so-call meta-paths. Our experiments show that GTNs learn new graph structures, based on data and tasks without domain knowledge, and yield powerful node representation via convolution on the new graphs. Without domain-specific graph preprocessing, GTNs achieved the best performance in all three benchmark node classification tasks against the state-of-the-art methods that require pre-defined meta-paths from domain knowledge.

Visual relationships provide higher-level information of objects and their relations in an image - this enables a semantic understanding of the scene and helps downstream applications. Given a set of localized objects in some training data, visual relationship detection seeks to detect the most likely "relationship" between objects in a given image. While the specific objects may be well represented in training data, their relationships may still be infrequent. The empirical distribution obtained from seeing these relationships in a dataset does not model the underlying distribution well - a serious issue for most learning methods. In this work, we start from a simple multi-relational learning model, which in principle, offers a rich formalization for deriving a strong prior for learning visual relationships. While the inference problem for deriving the regularizer is challenging, our main technical contribution is to show how adapting recent results in numerical linear algebra lead to efficient algorithms for a factorization scheme that yields highly informative priors. The factorization provides sample size bounds for inference (under mild conditions) for the underlying [object, predicate, object] relationship learning task on its own and surprisingly outperforms (in some cases) existing methods even without utilizing visual features. Then, when integrated with an end-to-end architecture for visual relationship detection leveraging image data, we substantially improve the state-of-the-art.

Statistical machine learning models that operate on manifold-valued data are being extensively studied in vision, motivated by applications in activity recognition, feature tracking and medical imaging. While non-parametric methods have been relatively well studied in the literature, efficientformulations for parametric models (which may offer benefits in small sample size regimes) have only emerged recently. Sofar, manifold-valued regression models (such as geodesic regression) are restricted to the analysis of crosssectional data, i.e., the so-called “fixed effects” in statistics. But in most “longitudinal analysis” (e.g., when a participant provides multiple measurements, over time) the application offixed effects models is problematic. In an effort to answer this need, this paper generalizes non-linear mixed effects model to the regime where the response variable is manifold-valued, i.e., f : Rd→ M. We derive the underlying model and estimation schemes and demonstrate the immediate benefits such a model can provide - both for group level and individual level analysis - on longitudinal brain imaging data. The direct consequence of our results is that longitudinal analysis of manifold-valued measurements (especially, the symmetric positive definite manifold) can be conducted in a computationally tractable manner.

Multi-label classification is the task of predicting a set of labels for a given input instance. Classifier chains are a state-of-the-art method for tackling such problems, which essentially converts this problem into a sequential prediction problem, where the labels are first ordered in an arbitrary fashion, and the task is to predict a sequence of binary values for these labels. In this paper, we replace classifier chains with recurrent neural networks, a sequence-to-sequence prediction algorithm which has recently been successfully applied to sequential prediction tasks in many domains. The key advantage of this approach is that it allows to focus on the prediction of the positive labels only, a much smaller set than the full set of possible labels. Moreover, parameter sharing across all classifiers allows to better exploit information of previous decisions. As both, classifier chains and recurrent neural networks depend on a fixed ordering of the labels, which is typically not part of a multi-label problem specification, we also compare different ways of ordering the label set, and give some recommendations on suitable ordering strategies.

A major goal of imaging studies such as the (ongoing) Human Connectome Project (HCP) is to characterize the structural network map of the human brain and identify its associations with covariates such as genotype, risk factors, and so on that correspond to an individual. But the set of image derived measures and the set of covariates are both large, so we must first estimate a 'parsimonious' set of relations between the measurements. For instance, a Gaussian graphical model will show conditional independences between the random variables, which can then be used to setup specific downstream analyses. But most such data involve a large list of 'latent' variables that remain unobserved, yet affect the 'observed' variables sustantially. Accounting for such latent variables is not directly addressed by standard precision matrix estimation, and is tackled via highly specialized optimization methods. This paper offers a unique harmonic analysis view of this problem. By casting the estimation of the precision matrix in terms of a composition of low-frequency latent variables and high-frequency sparse terms, we show how the problem can be formulated using a new wavelet-type expansion in non-Euclidean spaces. Our formulation poses the estimation problem in the frequency space and shows how it can be solved by a simple sub-gradient scheme. We provide a set of scientific results on ~500 scans from the recently released HCP data where our algorithm recovers highly interpretable and sparse conditional dependencies between brain connectivity pathways and well-known covariates.

Probability density functions (PDFs) are fundamental objects in mathematics with numerous applications in computer vision, machine learning and medical imaging. The feasibility of basic operations such as computing the distance between two PDFs and estimating a mean of a set of PDFs is a direct function of the representation we choose to work with. In this paper, we study the Gaussian mixture model (GMM) representation of the PDFs motivated by its numerous attractive features. (1) GMMs are arguably more interpretable than, say, square root parameterizations (2) the model complexity can be explicitly controlled by the number of components and (3) they are already widely used in many applications. The main contributions of this paper are numerical algorithms to enable basic operations on such objects that strictly respect their underlying geometry. For instance, when operating with a set of K component GMMs, a first order expectation is that the result of simple operations like interpolation and averaging should provide an object that is also a K component GMM. The literature provides very little guidance on enforcing such requirements systematically. It turns out that these tasks are important internal modules for analysis and processing of a field of ensemble average propagators (EAPs), common in diffusion weighted magnetic resonance imaging. We provide proof of principle experiments showing how the proposed algorithms for interpolation can facilitate statistical analysis of such data, essential to many neuroimaging studies. Separately, we also derive interesting connections of our algorithm with functional spaces of Gaussians, that may be of independent interest.

Statistical models for manifold-valued data permit capturing the intrinsic nature of the curved spaces in which the data lie and have been a topic of research for several decades. Typically, these formulations use geodesic curves and distances defined locally for most cases - this makes it hard to design parametric models globally on smooth manifolds. Thus, most (manifold specific) parametric models available today assume that the data lie in a small neighborhood on the manifold. To address this ’locality’ problem, we propose a novel nonparametric model which unifies multivariate general linear models (MGLMs) using multiple tangent spaces. Our framework generalizes existing work on (both Euclidean and non-Euclidean) general linear models providing a recipe to globally extend the locally-defined parametric models (using a mixture of local models). By grouping observations into sub-populations at multiple tangent spaces, our method provides insights into the hidden structure (geodesic relationships) in the data. This yields a framework to group observations and discover geodesic relationships between covariates X and manifold-valued responses Y, which we call Dirichlet process mixtures of multivariate general linear models (DP-MGLM) on Riemannian manifolds. Finally, we present proof of concept experiments to validate our model.

Linear regression is a parametric model which is ubiquitous in scientific analysis. The classical setup where the observations and responses, i.e., (xi, yi) pairs, are Euclidean is well studied. The setting where yi is manifold valued is a topic of much interest, motivated by applications in shape analysis, topic modeling, and medical imaging. Recent work gives strategies for max-margin classifiers, principal components analysis, and dictionary learning on certain types of manifolds. For parametric regression specifically, results within the last year provide mechanisms to regress one real-valued parameter, xi∈ R, against a manifold-valued variable, yi∈ M. We seek to substantially extend the operating range of such methods by deriving schemes for multivariate multiple linear regression -- a manifold-valued dependent variable against multiple independent variables, i.e., f: ℝn→ M. Our variational algorithm efficiently solves for multiple geodesic bases on the manifold concurrently via gradient updates. This allows us to answer questions such as: what is the relationship of the measurement at voxel y to disease when conditioned on age and gender. We show applications to statistical analysis of diffusion weighted images, which give rise to regression tasks on the manifold GL(n)/O(n) for diffusion tensor images (DTI) and the Hilbert unit sphere for orientation distribution functions (ODF) from high angular resolution acquisition. The companion open-source code is available on nitrc.org/projects/riem_mglm.