AAAI 2026 Papers — Page 40
AAAI Conference on Artificial Intelligence · 4149 papers
UDCH: Unsupervised Dynamic Weighted Cluster-cooperative Hashing for Cross-modal Retreival
Yuanzhi Zhao (Nanjing University of Finance and Economics), Xiaoyu Li (Nanjing University of Finance and Economics)
RetrievalContrastive LearningMultimodality
🎯 What it does: Propose an unsupervised dynamic weighted clustering collaborative hashing framework UDCH, which jointly learns cross-modal hash codes by integrating instance-level contrast, cluster-level semantic modeling, and structural regression.
UI-R1: Enhancing Efficient Action Prediction of GUI Agents by Reinforcement Learning
Zhengxi Lu (Zhejiang University), Hongsheng Li (Zhejiang University)
Computational EfficiencyLarge Language ModelReinforcement LearningVision-Language-Action ModelMultimodality
🎯 What it does: Propose the UI-R1 framework, leveraging rule-based reinforcement learning to enhance the reasoning and execution capabilities of multimodal large language models in GUI action prediction tasks, and design a two-phase efficient inference scheme called UI-R1-E.
UltraGen: High-Resolution Video Generation with Hierarchical Attention
Teng Hu (Shanghai Jiao Tong University), Ran Yi (Shanghai Jiao Tong University)
GenerationSuper ResolutionTransformerDiffusion modelVideo
🎯 What it does: Proposes the UltraGen framework, enabling local high-definition video generation from 1080P to 4K;
Ultralight Polarity-Split Neuromorphic SNN for Event-Stream Super-Resolution
Chuanzhi Xu (University of Sydney), Qiang Qu (University of Sydney)
Super ResolutionSpiking Neural Network
🎯 What it does: Proposed an ultra-lightweight, event-stream-to-event-stream super-resolution method, achieving real-time inference based on spiking neural networks (SNN).
UM-Text: A Unified Multimodal Model for Image Understanding and Visual Text Editing
Lichen Ma (Sun Yat-sen University), Junshi Huang (Sun Yat-sen University)
RecognitionImage TranslationGenerationLarge Language ModelVision Language ModelVision-Language-Action ModelDiffusion modelFlow-based ModelAuto EncoderImageTextMultimodalityBenchmark
🎯 What it does: Proposed the UM-Text unified multimodal model, enabling visual text editing and generation through natural language instructions.
UMNet: Uncertainty-guided Memory Network for Hyperspectral Pansharpening
Xiaozheng Wang (Tiangong University), Ziyang Liu (Tiangong University)
RestorationSuper ResolutionConvolutional Neural NetworkRecurrent Neural NetworkImage
🎯 What it does: Propose UMNet, a hyperspectral image fusion network based on a dual U-Net backbone, spatial-spectral recursive fusion unit (SRFU), and non-negative matrix factorization memory interaction unit (SMIU), achieving high spatial and spectral fidelity through two-stage uncertainty-guided loss.
Unaligned UAV RGBT Tracking: A Largescale Benchmark and a Novel Approach
Yun Xiao (Anhui University), Chenglong Li (Anhui University)
Object TrackingConvolutional Neural NetworkMixture of ExpertsVideoMultimodalityBenchmark
🎯 What it does: Propose the unaligned RGB-T tracking task for drones, construct the first large-scale unaligned RGB-T video dataset LUART, and design the SFCATrack method;
Unbiased Rectification for Sequential Recommender Systems Under Fake Orders
Qiyu Qin (Huazhong University of Science and Technology), Ruixuan Li (Huazhong University of Science and Technology)
Recommendation SystemLarge Language ModelContrastive LearningSequential
🎯 What it does: This paper proposes the DITaR framework for unbiased correction of fake orders embedded in real user sequences in sequential recommendation systems.
Uncertainty Quantification for Machine Learning: One Size Does Not Fit All
Paul Hofman (LMU Munich), Eyke Hüllermeier
Anomaly DetectionExplainability and InterpretabilityConvolutional Neural NetworkImageBiomedical Data
🎯 What it does: A second-order uncertainty measurement framework is proposed, which aligns with task loss for different downstream tasks (selective prediction, outlier detection, active learning), and is theoretically proven and experimentally validated.
Uncertainty-Based Methods for Automated Process Reward Data Construction and Output Aggregation in Mathematical Reasoning
Jiuzhou Han (Monash University), Ehsan Shareghi (Monash University)
Data-Centric LearningReinforcement Learning from Human FeedbackTransformerLarge Language ModelSupervised Fine-TuningReinforcement LearningText
🎯 What it does: Propose an uncertainty-driven process reward model (PRM) data construction and annotation framework, and design two uncertainty-aware aggregation methods (Hybrid Majority Reward Vote and Weighted Reward-Frequency Vote) that integrate Majority Vote with PRM for mathematical reasoning.
Uncertainty-Guided View-Strength-Aware Feature Utilization for Multi-View Classification
Li Lv (Shanxi University), Xinyan Liang (Taiyuan University of Science and Technology)
Classification
🎯 What it does: Designed and implemented the UVF framework, which guides view-intensity-aware feature utilization and fusion through uncertainty estimation, achieving multi-view classification.
Uncertainty-Propelled Physics-MAE Fusion for Self-Supervised Diffusion-Weighted Image Denoising
Zeyu Deng (Guizhou University), Yingfeng Ou (Guizhou University)
RestorationDiffusion modelAuto EncoderBiomedical DataMagnetic Resonance ImagingDiffusion Tensor Imaging
🎯 What it does: Propose a UP-MAE fusion framework combining MAE and physics-guided Noise2Noise for self-supervised diffusion-weighted image denoising.
Uncovering and Mitigating Destructive Multi-Embedding Attacks in Deepfake Proactive Forensics
Lixin Jia (Xinjiang University), Gaobo Yang (Hefei University of Technology)
Anomaly DetectionSafty and PrivacyAdversarial AttackImage
🎯 What it does: This paper first reveals the vulnerability of multi-embedding attacks (MEA) in active deepfake forensics and proposes an adversarial interference simulation (AIS) training paradigm to enhance watermark robustness.
Uncovering and Mitigating Transient Blindness in Multimodal Model Editing
XiaoQi Han, Jeff Z. Pan (Shanxi University)
Explainability and InterpretabilityAdversarial AttackData-Centric LearningVision Language ModelMultimodalityBenchmark
🎯 What it does: This paper proposes a dynamic evaluation framework called De-VQA for multi-modal model editing (MMED), utilizing three locality dimensions (random image, no image, and consistent image, totaling seven categories of data) to systematically detect and quantify the model's dependency on visual information after editing, identifying and mitigating the 'Transient Blindness' phenomenon.
Uncovering Hidden Degeneration: A Physics-Guided Bidirectional Inference Framework for Industrial Time Series Prediction
Xingwang Li (Southwest Jiaotong University), Qiang Duan (Pennsylvania State University)
Anomaly DetectionTransformerTime SeriesPhysics Related
🎯 What it does: Propose a physics-guided bidirectional reasoning framework that combines continuous damage mechanics simulation with maximum entropy inference to predict hidden degradation processes in industrial time series.
Uncovering Pretraining Code in LLMs: A Syntax-Aware Attribution Approach
Yuanheng Li (Tongji University), Shengjie Zhao (Tongji University)
Adversarial AttackLarge Language ModelText
🎯 What it does: Developed a membership inference attack method for code called SYNPRUNE, which improves the detection accuracy of code samples in LLM training data by removing tokens that are necessarily present under syntactic conventions.
Under-Approximating Semantics in Clustered Assumption-Based Argumentation
Iosif Apostolakis (Graz University of Technology), Johannes P. Wallner (Graz University of Technology)
Computational EfficiencyBiomedical Data
🎯 What it does: Propose clustering abstraction on questionable parts within Assumption-Based Argumentation (ABA) to simplify complex scenarios while ensuring reasoning safety.
Understanding and Enhancing Differentiable Architecture Search from Information Bottleneck Perspective
Haidong Kang, Bo Yi (Northeastern University)
Neural Architecture SearchImageBenchmark
🎯 What it does: Proposed a Batch Entropy-decay Regularization (BER) method based on the information bottleneck theory to help differential architecture search (DAS) avoid performance collapse.
Understanding Dynamic Scenes in Ego Centric 4D Point Clouds
Junsheng Huang (Zhejiang University), Gaoang Wang (Zhejiang University)
Object DetectionObject TrackingSegmentationTransformerLarge Language ModelSupervised Fine-TuningVision Language ModelVideoMultimodalityPoint CloudBenchmarkChain-of-Thought
🎯 What it does: This paper proposes a four-dimensional dynamic scene question answering benchmark called EgoDynamic4D for first-person perspective, and designs an end-to-end spatiotemporal reasoning framework;
Understanding Interaction as You Need: Intention-Driven Pedestrian Behavior Prediction
Hang Yu (Shanghai University), Jiayan Qiu (University of Leicester)
Autonomous DrivingExplainability and InterpretabilityConvolutional Neural NetworkRecurrent Neural NetworkTransformerPoint CloudSequential
🎯 What it does: Propose an intent-driven interaction modeling framework for simultaneously predicting pedestrian trajectory and crossing intent.
Understanding the Impact of Proportionality in Approval-Based Multiwinner Elections
Niclas Boehmer (Hasso Plattner Institute, University of Potsdam), Jannik Peters (National University of Singapore)
Tabular
🎯 What it does: This paper studies the practical impact of the proportionality axiom in approval-based multi-choice committee elections, combining algorithmic analysis with experimental evaluation to assess its constraints on selectable committees and candidate importance.
UNeMo: Collaborative Visual-Language Reasoning and Navigation via a Multimodal World Model
Changxin Huang (Shenzhen University), Jianqiang Li (Shenzhen University)
Large Language ModelVision-Language-Action ModelAuto EncoderWorld ModelMultimodalityBenchmark
🎯 What it does: Proposes the UNeMo framework, which achieves collaborative optimization of visual state reasoning and navigation decision-making through a multi-modal world model and a hierarchical prediction-feedback mechanism.
UniABG: Unified Adversarial View Bridging and Graph Correspondence for Unsupervised Cross-View Geo-Localization
Cuiqun Chen (Anhui University), Xingyi Zhang (Anhui University)
RetrievalDomain AdaptationConvolutional Neural NetworkGraph Neural NetworkGenerative Adversarial NetworkContrastive LearningImage
🎯 What it does: Propose a two-stage unsupervised cross-view geolocation framework named UniABG, first eliminating perspective differences through perspective-aware adversarial bridging, then improving cross-view correspondences via heterogeneous graph filtering correction.
UniAlignment: Semantic Alignment for Unified Image Generation, Understanding, Manipulation and Perception
Xinyang Song (University of Chinese Academy of Sciences), Zhenan Sun (University of Chinese Academy of Sciences)
GenerationTransformerDiffusion modelAuto EncoderContrastive LearningImageTextMultimodalityBenchmark
🎯 What it does: Propose UniAlignment, a unified multi-modal generation framework based on a single Diffusion Transformer, capable of simultaneously performing image generation, understanding, editing, perception, and personalization tasks.
UniAPO: Unified Multimodal Automated Prompt Optimization
Qipeng Zhu (ByteDance Inc), Zhenheng Yang (ByteDance Inc)
OptimizationTransformerLarge Language ModelPrompt EngineeringImageVideoTextMultimodalityRetrieval-Augmented GenerationChain-of-Thought
🎯 What it does: For multimodal tasks, the UniAPO framework is proposed, which decouples feedback modeling from prompt optimization through an EM-inspired optimization process, and enhances the effectiveness of automatic prompt optimization by incorporating long- and short-term memory mechanisms and process-level supervision.
UniC-Lift: Unified 3D Instance Segmentation via Contrastive Learning
Ankit Dhiman (Indian Institute of Science), Venkatesh Babu Radhakrishnan (Samsung R&D Institute India)
SegmentationContrastive LearningGaussian SplattingImage
🎯 What it does: To address the problem of poor 3D instance segmentation caused by inconsistent 2D instance segmentation labels across multiple views, this paper proposes a unified framework called UniC-Lift. It introduces learnable vector embeddings for each Gaussian atom in 3D Gaussian Splatting (3DGS), achieves consistent segmentation through alignment, triplet loss, and spatial smoothing regularization, and finally obtains 3D instance labels directly via a simple 'Embedding-to-Label' decoding process.
UniCUE: Unified Recognition and Generation Framework for Chinese Cued Speech Video-to-Speech Generation
Jinting Wang (Hong Kong University of Science and Technology), Li Liu (Hong Kong University of Science and Technology)
RecognitionGenerationTransformerDiffusion modelContrastive LearningVideoMultimodalityAudio
🎯 What it does: Proposed the UniCUE framework to directly generate speech from Chinese sign language videos, integrating visual understanding with speech generation.
Unified Interaction Consistency Learning for Single-Source Domain-Generalized Object Detection in Urban Scene
Peng Zhang (Northwestern Polytechnical University), Gong Cheng (Northwestern Polytechnical University)
Object DetectionDomain AdaptationAutonomous DrivingKnowledge DistillationConvolutional Neural NetworkContrastive LearningImageBenchmark
🎯 What it does: Proposed a Unified Interactive Consistent Learning (UICL) framework to address object detection under single-source domain generalization.
Unified Minimax Optimization Framework for Propensity Score Estimation in Debiased Recommendation
Chunyuan Zheng (Peking University), Tianyu Xia (Peking University)
Recommendation SystemOptimization
🎯 What it does: Proposed a unified minimax optimization framework called UPO for jointly estimating and correcting propensity scores in recommendation systems to eliminate bias caused by MNAR.
Unified Mixture-of-Experts Framework for Joint Cardiac and Vascular Ultrasound Analysis and Report Generation
Bin Pu (Hunan University), Kenli Li (Hunan University)
GenerationExplainability and InterpretabilityLarge Language ModelMixture of ExpertsImageTextMultimodalityBiomedical DataUltrasound
🎯 What it does: This paper proposes the ECV framework, achieving joint parameter measurement and report generation for cardiac and vascular ultrasound images.
Unified Representation Causal Prompt Distillation for Re-Inference-Free Lifelong Person Re-Identification
Jiaqi Zhao (China University of Mining and Techology), Rui Yao (China University of Mining and Techology)
RecognitionKnowledge DistillationRepresentation LearningTransformerPrompt EngineeringImage
🎯 What it does: Propose a zero-shot inference lifelong person re-identification method named Unified Representation Causal Prompt Distillation (URCPD), combining feature decoupled style transfer and causal prompt distillation, which enhances cross-domain generalization while significantly alleviating catastrophic forgetting.
Unified Structural Factors for Transfer Learning Generalization with PAC-Bayesian Guarantees
Ziqi Gao (University of Pennsylvania)
Domain AdaptationRepresentation LearningTransformerText
🎯 What it does: Propose a unified structured framework based on Feature Overlap Rate (FOR) and Effective Task Complexity (ETC) to explain the generalization performance of transfer learning;
Unified View Extraction with Low-Rankness and Smoothness Fusion for Multi-View Subspace Clustering
Yapeng Wang, Ming Yang (Xidian University)
OptimizationRepresentation LearningMultimodality
🎯 What it does: Propose a unified multi-view subspace clustering method called UVELRS, which first extracts consistent cross-view representations from self-representation matrices of each view, and then constructs a tensor and jointly optimizes it via tensor total variation Schatten p norm.
UniFit: Towards Universal Virtual Try-on with MLLM-Guided Semantic Alignment
Wei Zhang (Nanjing University of Science and Technology), Zhichao Lian (Southeast University)
Image TranslationImage HarmonizationGenerationData SynthesisPose EstimationTransformerLarge Language ModelDiffusion modelAuto EncoderImageTextMultimodality
🎯 What it does: Propose a general virtual try-on framework called UniFit, guided by a multi-modal large language model (MLLM), which can perform various try-on tasks through text instructions.
Unifying Channel Independence and Mixing: Multi-Scale Patch Recursion for Global–Local Representation Synergy in Multivariate Time Series Forecasting
Wenhao Zhang (Beijing Jiaotong University), Shaoxiong Pang (Beijing Jiaotong University)
Computational EfficiencyRepresentation LearningTime Series
🎯 What it does: Proposed a pure MLP-driven multi-scale patch recursive framework named FusionTimePatch for multivariate time series forecasting.
Unifying Locality of KANs and Feature Drift Compensation Projection for Data-Free Replay Based Continual Face Forgery Detection
Tianshuo Zhang (University of Chinese Academy of Sciences), Zhen Lei (Chinese Academy of Sciences)
ClassificationAnomaly DetectionKnowledge DistillationContrastive LearningImage
🎯 What it does: Propose a continuity-based human face forgery detection framework KAN-CFD based on Kolmogorov-Arnold Networks (KAN), addressing the catastrophic forgetting problem of traditional methods when learning new tasks.
Unifying Multi-View Knowledge for Graph Learning via Model Collaboration
Zhihao Wu (Zhejiang University), Haishuai Wang (Zhejiang University)
Computational EfficiencyRepresentation LearningGraph Neural NetworkTransformerLarge Language ModelSupervised Fine-TuningMixture of ExpertsTextGraph
🎯 What it does: Propose the COLA framework in text attribute graph learning, achieving multi-model collaborative learning by integrating frozen LLMs, fine-tuned SLMs, and GNNs;
UniHOI: Unified Human-Object Interaction Understanding via Unified Token Space
Panqi Yang (Xi'an Jiao Tong University), Yongqiang Ma (Xi'an Jiao Tong University)
Object DetectionGenerationTransformerLarge Language ModelVision Language ModelVision-Language-Action ModelAuto EncoderImageTextMultimodality
🎯 What it does: Propose UniHOI, unifying human-object interaction detection and image generation through a shared discrete vocabulary to enable bidirectional mapping, thereby enhancing interaction understanding and generation quality.
UniHR: Hierarchical Representation Learning for Unified Knowledge Graph Link Prediction
Zhiqiang Liu (Zhejiang University), Wen Zhang (Zhejiang University)
Representation LearningGraph Neural NetworkTransformerGraph
🎯 What it does: Proposes UniHR, a unified hierarchical knowledge graph representation learning framework that can convert three complex facts—hyper-relational, temporal, and nested—into triplet forms. It learns semantic and structural information of facts and entities through internal and external message passing, ultimately completing link prediction for multi-type knowledge graphs.
UniMapGen: A Generative Framework for Large-Scale Map Construction from Multi-modal Data
Yujian Yuan (Amap, Alibaba Group), Mu Xu (Amap, Alibaba Group)
GenerationAutonomous DrivingTransformerLarge Language ModelPrompt EngineeringImageTextMultimodality
🎯 What it does: Propose the UniMapGen generative framework to construct large-scale lane-level maps from multimodal data (BEV, PV, text prompts)
UniME-V2: MLLM-as-a-Judge for Universal Multimodal Embedding Learning
Tiancheng Gu (MiroMind AI), Lidong Bing (MiroMind AI)
RetrievalRepresentation LearningTransformerLarge Language ModelVision Language ModelContrastive LearningImageTextMultimodality
🎯 What it does: Propose a general multimodal embedding framework, UniME-V2, which leverages multimodal large language models (MLLM) for judgment, and design hard negative mining and distribution alignment training based on soft semantic matching scores, followed by the construction of UniME-V2-Reranker to enhance retrieval accuracy.
UniMGS: Unifying Mesh and 3D Gaussian Splatting with Single-Pass Rasterization and Proxy-Based Deformation
Zeyu Xiao (Fudan University), Lihua Zhang (Fudan University)
Gaussian SplattingPoint CloudMesh
🎯 What it does: Propose the UniMGS framework to achieve unified single-channel rendering and proxy-based deformation for meshes and 3D Gaussian Splatting (3DGS);
UniMM-V2X: MoE-Enhanced Multi-Level Fusion for End-to-End Cooperative Autonomous Driving
Ziyi Song (Tsinghua University), Zhisheng Niu (Tsinghua University)
Autonomous DrivingTransformerMixture of ExpertsMultimodality
🎯 What it does: Propose the UniMM-V2X framework to achieve end-to-end autonomous driving with multi-agent collaboration at perception and prediction levels.
UniMo: Unified Motion Generation and Understanding with Chain of Thought
Guocun Wang (Tsinghua University), Xiaoguang Han (Chinese University of Hong Kong)
GenerationTransformerLarge Language ModelReinforcement LearningVision Language ModelAuto EncoderVideoTextMultimodalityChain-of-Thought
🎯 What it does: Construct a unified model named UniMo that integrates text-to-motion (T2M) and motion-to-text (M2T) tasks, leveraging large language models combined with chain-of-thought reasoning and reinforcement learning to achieve complementary improvements in generation and understanding.
UniScene-MoTion: Unified Scene & Motion-aware Diffusion Transition Framework
Rui Jiang (Zhejiang University), Xi Li (Zhejiang University)
GenerationDepth EstimationTransformerDiffusion modelVideo
🎯 What it does: Proposed the unified scene and motion perception multi-scale video transition framework UniScene-Motion.
UniSketch: A Unified Framework for Parametric Sketch Generation and Constraint Prediction
Jing Lin, Rubin Fan (Wuhan University)
GenerationTransformerImageSequential
🎯 What it does: Proposes a unified framework capable of generating parametric sketches from images, predicting constraints, and performing unconditional sketch synthesis.
Universal Adversarial Purification with DDIM Metric Loss for Stable Diffusion
Li Zheng (University of Macau), He YiMin (University of Macau)
Adversarial AttackDiffusion modelScore-based ModelAuto EncoderImage
🎯 What it does: Propose a general adversarial purification framework UDAP for Stable Diffusion, utilizing DDIM inversion reconstruction error for adversarial noise elimination, and achieving efficient purification through dynamic epoch adjustment.
Universal Compressed Image Restoration via Codec-Aware Conditioning with Reinforcement Learning
Changwoo Han (Korea University), Seung-Won Jung (Electronics and Telecommunications Research Institute)
RestorationCompressionTransformerReinforcement LearningImage
🎯 What it does: Propose a unified UCIR framework, where P-Net predicts the encoder type and compression level, and this information is fed into R-Net for conditional image restoration.
Universal EEG Epilepsy Detection via Evidential Multi-View De-Biasing
Ziqi Wen (Xidian University), Wei Zhao (Northwest University)
ClassificationAnomaly DetectionBiomedical Data
🎯 What it does: Proposed a multi-view debiasing framework BF-EML based on evidence inference for achieving global seizure detection without retraining.
Universal Learning of Stochastic Dynamics for Exact Belief Propagation Using Bernstein Normalizing Flows
Peter Amorese (University of Colorado Boulder), Morteza Lahijanian (University of Colorado Boulder)
Representation LearningFlow-based ModelTime SeriesPhysics Related
🎯 What it does: Proposed a Bernstein Normalizing Flows (BNF) model to learn nonlinear stochastic dynamics and enable exact belief propagation.
Universal Safety Controllers with Learned Prophecies
Bernd Finkbeiner (CISPA Helmholtz Center for Information Security), Anne-Kathrin Schmuck (CISPA Helmholtz Center for Information Security)
Explainability and InterpretabilityComputational EfficiencyBenchmark
🎯 What it does: Proposes a learning-based method for constructing Universal Safety Controllers (USC), which learns CTL formulas from example plants to approximate prophecy, thereby generating generalizable, interpretable, and efficient controllers.
Unlearning in Cross-Modal Retrieval via Prior-Prototype Guided Partitioned Dampening
Yi Lu (Xinjiang University), Yurong Qian (Xinjiang University)
RetrievalMultimodality
🎯 What it does: Propose a no-training reset framework PPP for cross-modal retrieval, achieving selective forgetting for specific categories through prior prototype-guided parameter partitioning and hierarchical suppression.
Unleashing Semantic and Geometric Priors for 3D Scene Completion
Shiyuan Chen (Soochow University), Cong Yang (D-Robotics)
SegmentationDepth EstimationAutonomous DrivingTransformerGaussian SplattingImageMultimodalityPoint Cloud
🎯 What it does: Propose the FoundationSSC framework, achieving high-quality 3D scene completion by dual decoupling at the source and channel layers, leveraging foundation models to provide separated semantic and geometric priors.
Unleashing the Potential of Large Language Models for Text-to-Image Generation Through Autoregressive Representation Alignment
Xing Xie (Chinese Academy of Sciences), Liangqiong Qu (University of Hong Kong)
GenerationTransformerLarge Language ModelVision Language ModelImageBiomedical Data
🎯 What it does: Propose a training framework called ARRA that does not require modifying the architecture of large language models (LLMs), enabling LLMs to generate globally consistent images while maintaining the autoregressive paradigm.
Unleashing the Power of Image-Tabular Self-Supervised Learning via Breaking Cross-Tabular Barriers
Yibing Fu (National University of Singapore), Yueming Jin (National University of Singapore)
Convolutional Neural NetworkTransformerMixture of ExpertsContrastive LearningImageMultimodalityTabularBiomedical DataAlzheimer's Disease
🎯 What it does: Proposed the CITab framework to achieve cross-table image-table self-supervised learning, enhancing the transferability of medical multimodal models.
Unlocking Dynamic Inter-Client Spatial Dependencies: A Federated Spatio-temporal Graph Learning Method for Traffic Flow Forecasting
Feng Wang (Beijing Advanced Innovation Center for Future Blockchain and Privacy Computing), Zhiming Zheng (Beijing Advanced Innovation Center for Future Blockchain and Privacy Computing)
Federated LearningRepresentation LearningRecurrent Neural NetworkGraph Neural NetworkGraphTime Series
🎯 What it does: Designed and implemented the FedSTGD framework to address the modeling and reconstruction of dynamic cross-client spatial dependencies in a federated learning environment.
Unlocking Efficient Vehicle Dynamics Modeling via Analytic World Models
Asen Nachkov (INSAIT, Sofia University 'St. Kliment Ohridski'), Luc Van Gool (University of Zurich)
Autonomous DrivingOptimizationRecurrent Neural NetworkSupervised Fine-TuningReinforcement LearningWorld ModelVideoSequential
🎯 What it does: This paper proposes and implements three Analytic World Models (AWMs) based on differentiable simulation: relative odometry prediction, optimal planner, and inverse optimal state estimation, further expanding the application scope of differentiable simulation in autonomous driving.
Unlocking Multi-Modal Potentials for Link Prediction on Dynamic Text-Attributed Graphs
Yuanyuan Xu (University of New South Wales), Xiwei Xu (Shanghai Jiao Tong University)
Representation LearningGraph Neural NetworkTransformerLarge Language ModelContrastive LearningTextMultimodalityGraph
🎯 What it does: Propose the MoMent model for multimodal representation learning on dynamic text attributed graphs (DyTAG), which first independently extracts node features through three modalities (time, text, structure), then fuses them to generate final node vectors, and enforces modal consistency via a dual-domain alignment loss.
Unnoticed Yet Effective: A Hybrid Physical Camouflage Framework Against DNNs and Human Perception
Mingye Xie (Shanghai Jiao Tong University), Yuzhuo Fu (Shanghai Jiao Tong University)
Autonomous DrivingAdversarial AttackTransformerLarge Language ModelPrompt EngineeringImage
🎯 What it does: Propose the UYE framework to generate physical camouflage patterns that can both mislead deep neural networks and remain stealthy to human perception.
UNO! UNified Offline Training Paradigm for Learning Path Recommendation
Linzhi Peng (Beihang University), Weifeng Lv (Beihang University)
Recommendation SystemTransformerReinforcement LearningSequential
🎯 What it does: Propose the UNO framework, combining offline training and process rewards to provide personalized learning path recommendations
Unpacking the Implicit Norm Dynamics of Sharpness-Aware Minimization in Tensorized Models
Tianxiao Cao (Kyoto University), Hisashi Kashima (Kyoto University)
CompressionOptimizationComputational EfficiencyImageText
🎯 What it does: This paper investigates the implicit regularization behavior of Sharpness-Aware Minimization (SAM) in multi-core tensor models and proposes an explicit regularization method called Deviation-Aware Scaling (DAS) based on core norm deviation.
Unreal-MAP: Unreal-Engine-Based General Platform for Multi-agent Reinforcement Learning
Tianyi Hu (Institute of Automation, Chinese Academy of Sciences), Tenghai Qiu (Institute of Automation, Chinese Academy of Sciences)
Robotic IntelligenceReinforcement LearningBenchmark
🎯 What it does: Developed a general-purpose multi-agent reinforcement learning (MARL) platform called Unreal-MAP based on Unreal Engine, and built an experimental framework that can seamlessly integrate with various algorithm frameworks (e.g., HMAP, PyMARL, HARL). It provides a five-layer modular architecture allowing customizable tasks, maps, teams, entities, and events.
UNSEEN: Enhancing Dataset Pruning from a Generalization Perspective
Furui Xu (Shanghai Jiao Tong University), Linfeng Zhang (Beijing Jiaotong University)
ClassificationData-Centric LearningConvolutional Neural NetworkImage
🎯 What it does: Propose a dataset pruning framework called UNSEEN based on a generalization perspective, and implement an incremental selection (IS) strategy on it to dynamically build a high-quality coreset.
Unsupervised Combinatorial Probabilistic Reasoning: Probabilistic Coin Change Problem
Zhongdi Qu (Cornell University), Carla P. Gomes (Cornell University)
Representation LearningRecurrent Neural NetworkBiomedical Data
🎯 What it does: Studied the Probabilistic Coin Change Problem (PCCP) and proposed an unsupervised DeepProReasoner framework
Unsupervised Contrastive Learning for Efficient and Robust Spectral Shape Matching
Feifan Luo (Zhejiang University), Hongyang Chen (Zhejiang University)
Computational EfficiencyRepresentation LearningGraph Neural NetworkContrastive LearningMesh
🎯 What it does: Proposed an unsupervised contrastive learning-based deep functional map network for efficient and robust non-rigid 3D shape matching.
Unsupervised Feature Selection Through Group Discovery
Shira Lifshitz (Technion Israel Institute of Technology), Hadas Benisty (Technion Israel Institute of Technology)
Representation LearningImageTabularBiomedical Data
🎯 What it does: Propose a fully unsupervised, differentiable end-to-end framework called GroupFS to simultaneously learn feature group assignments and select the most important feature groups, achieving feature selection for high-dimensional data.
Unsupervised Motion-Compensated Decomposition for Cardiac MRI Reconstruction via Neural Representation
Xuanyu Tian (ShanghaiTech University), Hongjiang Wei (University of Michigan)
RestorationConvolutional Neural NetworkNeural Radiance FieldOptical FlowBiomedical DataMagnetic Resonance Imaging
🎯 What it does: Propose an unsupervised motion-compensated implicit neural representation (MoCo-INR) framework for reconstructing high-quality cardiac MRI images from highly undersampled k-t space data.
Unsupervised Multi-Parameter Inverse Solving for Reducing Ring Artifacts in 3D X-Ray CBCT
Qing Wu (ShanghaiTech University), Yuyao Zhang (ShanghaiTech University)
RestorationBiomedical DataComputed Tomography
🎯 What it does: Riner proposed an unsupervised multi-parameter inverse solution method that directly estimates the detector's non-ideal response from raw projection data and reconstructs 3D CBCT images without ring artifacts.
Unsupervised Multi-View Visual Anomaly Detection via Progressive Homography-Guided Alignment
Xintao Chen (ShanghaiTech University), Yingna Wu (ShanghaiTech University)
Anomaly DetectionDiffusion modelImage
🎯 What it does: Proposed an unsupervised multi-view visual anomaly detection framework called ViewSense-AD (VSAD), which achieves viewpoint-invariant features through layer-wise alignment based on disparity alignment;
Unsupervised Semantic Discovery via Global and Local Semantic Alignment in Multimodal Clustering
Zhengzhong Zhu (Sichuan University), Jiangping Zhu (Sichuan University)
Representation LearningTransformerContrastive LearningMultimodality
🎯 What it does: This paper proposes the GLAD framework to achieve unsupervised multimodal semantic discovery, improving clustering quality through global and local semantic alignment.
Unsupervised Single-Channel Audio Separation with Diffusion Source Priors
Runwu Shi (Institute of Science Tokyo), Kazuhiro Nakadai (Institute of Science Tokyo)
RestorationTransformerDiffusion modelScore-based ModelAudio
🎯 What it does: Using unsupervised diffusion models as source priors, combined with gradient-guided inverse problem solving, to achieve single-channel audio source separation.
Unveiling the Attribute Misbinding Threat in Identity-Preserving Models
Junming Fu (Sun Yat-sen University), Jianquan Yang (Sun Yat-sen University)
GenerationSafty and PrivacyAdversarial AttackTransformerLarge Language ModelPrompt EngineeringDiffusion modelImageTextMultimodality
🎯 What it does: This paper proposes the Attribute Misbinding Attack, which exploits attention bias in text-to-image diffusion models to cause the model to incorrectly bind sensitive attributes to target identities, thereby generating NSFW images.
Unveiling the Fragility of Vision-Language Models: Multi-Modal Adversarial Synergy via Texture-Constrained Perturbations and Cross-Modal Optimization
Xiang Fang (Huazhong University of Science and Technology), Changshuo Wang (Huazhong University of Science and Technology)
OptimizationAdversarial AttackPrompt EngineeringVision Language ModelImageTextMultimodality
🎯 What it does: Propose an MMAS framework that can perform unified, cross-task adversarial attacks on large vision-language models under black-box conditions by leveraging texture-constrained general visual perturbations and learnable text perturbations.
Unveiling the Landscape of Clinical Depression Assessment: From Behavioral Signatures to Psychiatric Reasoning
Zhuang Chen, Minlie Huang (Tsinghua University)
Convolutional Neural NetworkRecurrent Neural NetworkLarge Language ModelVideoTextMultimodalityBiomedical DataAudio
🎯 What it does: Constructed and publicly released the C-MIND clinical multimodal depression diagnosis dataset, and systematically evaluated its behavioral characteristics and diagnostic value.
Uplift Modeling with Delayed Feedback: Identifiability and Algorithms
Chunyuan Zheng (Peking University), Zhouchen Lin (Peking University)
Tabular
🎯 What it does: Studies methods for estimating uplift modeling under delayed feedback scenarios, proposing a joint framework that simultaneously learns latent response times and potential outcomes.
UQ-Bench: A Benchmark for Evaluating Multimodal LLMs on Underwater Image Quality Assessment
Jingchao Cao (Ocean University of China), Yutao Liu (Ocean University of China)
TransformerLarge Language ModelPrompt EngineeringVision Language ModelImageTextMultimodalityBenchmark
🎯 What it does: This paper designs the UQ-Bench benchmark to systematically evaluate the capability of multimodal large language models in low-level visual perception and quality assessment of underwater images.
UQ-ViT: Harmonizing Extreme Activations with Hardware-Friendly Uniform Quantization in Vision Transformers
Tao Jiang (Northwestern Polytechnical University), Junwei Han (Northwestern Polytechnical University)
ClassificationObject DetectionSegmentationComputational EfficiencyTransformerImage
🎯 What it does: This work proposes UQ-ViT, a unified quantization framework designed to handle extreme activation distributions in Vision Transformers, achieving hardware-friendly and high-precision low-bit quantization.
URaG: Unified Retrieval and Generation in Multimodal LLMs for Efficient Long Document Understanding
Yongxin Shi (South China University of Technology), Lianwen Jin (South China University of Technology)
GenerationRetrievalComputational EfficiencyTransformerLarge Language ModelSupervised Fine-TuningVision Language ModelTextMultimodalityBenchmarkRetrieval-Augmented Generation
🎯 What it does: Propose a unified retrieval and generation framework, URaG, which realizes document retrieval and answer generation within multimodal large language models (LLMs), addressing issues of information interference and computational costs in long document understanding.
UrbanNav: Learning Language-Guided Embodied Urban Navigation from Web-Scale Human Trajectories
Yanghong Mei (Institute of Automation Chinese Academy of Sciences), Jing Liu (Institute of Automation Chinese Academy of Sciences)
Autonomous DrivingRobotic IntelligenceTransformerVision Language ModelVision-Language-Action ModelSimultaneous Localization and MappingMultimodality
🎯 What it does: Built an extensible framework called UrbanNav for unsupervised language-instructed urban navigation training using robot-compatible trajectories and natural language instructions automatically extracted from web-scale human walking videos.
UrbanPG: An Efficient Framework with Personalized Context and General Backbone Interaction for Urban Spatio-Temporal Learning
Aoyu Liu (Tongji University), Yaying Zhang (Tongji University)
OptimizationComputational EfficiencyRepresentation LearningTransformerPrompt EngineeringGraphTime Series
🎯 What it does: Propose the UrbanPG framework, decomposing urban spatiotemporal prediction into a general backbone and personalized context prompts to enhance performance in large-scale, few-shot, and continual learning scenarios.
URPO: A Unified Reward & Policy Optimization Framework for Large Language Models
Songshuo Lu (Moore Threads AI), Yaohua Tang (Moore Threads AI)
OptimizationReinforcement Learning from Human FeedbackTransformerLarge Language ModelReinforcement LearningText
🎯 What it does: This study proposes a unified reward and policy optimization framework, URPO, which employs a single large model to simultaneously assume the roles of generator (player) and evaluator (referee). It integrates rule-verified reasoning, open-ended instructions, and preference data within a single GRPO training loop, achieving an integrated reinforcement learning alignment process.
USE: A Unified Model for Universal Sound Separation and Extraction
Hongyu Wang (Shanghai Jiao Tong University), Yanmin Qian (Shanghai Jiao Tong University)
Convolutional Neural NetworkRecurrent Neural NetworkTransformerContrastive LearningVideoTextMultimodalityAudio
🎯 What it does: Proposes a unified USE framework capable of performing global source separation (SS) when the number of unknown sound sources is unknown, or target source extraction (TSE) when multimodal cues are provided
Using Certifying Constraint Solvers for Generating Step-wise Explanations
Ignace Bleukx (KU Leuven), Tias Guns (KU Leuven)
OptimizationExplainability and InterpretabilityComputational Efficiency
🎯 What it does: Proposed a method to rapidly construct step-by-step explanations using proofs generated by evidence constraint solvers, significantly reducing explanation generation time;
Using Constraint Solvers to Construct Binary Codes with Good Error Correction Performance
Stepan Kochemazov (ITMO University), Alexander Semenov (ITMO University)
OptimizationBenchmark
🎯 What it does: Construct binary linear codes with excellent error correction performance using constraint solvers.
USPR: Learning a Unified Solver for Profiled Routing
Chuanbo Hua (KAIST), Jinkyoo Park (KAIST)
OptimizationReinforcement Learning from Human FeedbackTransformerReinforcement LearningTabularBenchmark
🎯 What it does: Proposed a unified learning-based solver, USPR, for handling the Profiled Vehicle Routing Problem (PVRP) that includes vehicle-customer preferences and constraints;
Utilitarian Guarantees for the Method of Equal Shares
Anton Baychkov (University of Warwick), Jannik Peters (National University of Singapore)
Optimization
🎯 What it does: This paper studies the utility guarantees of the Method of Equal Shares (MES) in participatory budgeting (PB) under fairness rules, providing the minimum utility ratio that MES can ensure under the DNS (weakly decreasing normalized) satisfaction function when parameterized by budget, minimum/maximum project costs. It proves that this bound is tight in the limit; meanwhile, it also provides corresponding guarantees for the GREEDY rule and scenarios using incorrect satisfaction functions. Additionally, it presents negative results showing that functions other than DNS lead to meaningless guarantees.
UV-RGS: Relightable 3D Gaussian Splatting from Unposed Views Under Varied Illuminations
Wei Feng (Tianjin University), Nan Li (Tianjin University)
GenerationOptimizationGaussian SplattingImagePoint Cloud
🎯 What it does: Proposes a framework named UV-RGS that learns a relightable 3D Gaussian Splatting model using uncalibrated camera poses and views with varying illumination, enabling scene inverse rendering and relighting.
UVLM: Benchmarking Video Language Model for Underwater World Understanding
Xizhe Xue (Northwestern Polytechnical University), Rong Xiao (Intellifusion Inc)
ClassificationTransformerLarge Language ModelSupervised Fine-TuningVision Language ModelVideoTextMultimodalityBenchmark
🎯 What it does: Built the first underwater multimodal video-language benchmark UVLM, covering 419 marine species, 0.9 million frames, 20 tasks, using human-AI collaborative annotation and GPT-4o-assisted text generation.
V-Pruner: A Fast and Globally-informed Token Pruning Framework for Vision Transformer
Guangzhen Yao (Northeast Normal University), Runhao Liu (Northeast Normal University)
ClassificationComputational EfficiencyTransformerReinforcement LearningImage
🎯 What it does: V-Pruner first evaluates token importance using Fisher information, then applies Proximal Policy Optimization (PPO) reinforcement learning to perform globally sequential token pruning on Vision Transformers, significantly reducing FLOPs, inference latency, while maintaining high accuracy.
V2VLoc: Robust GNSS-Free Collaborative Perception via LiDAR Localization
Wenkai Lin (Xiamen University), Chenglu Wen (Xiamen University)
Object DetectionPose EstimationAutonomous DrivingTransformerPoint Cloud
🎯 What it does: This paper proposes a GNSS-free vehicle-to-vehicle collaborative perception framework based on LiDAR localization, using PGC to estimate pose and confidence, and then using PASTAT to perform spatiotemporal adaptive alignment of features, achieving multi-vehicle collaborative detection;
VaccineRAG: Boosting Multimodal Large Language Models’ Immunity to Harmful RAG Samples
Qixin Sun (Beihang University), Linjiang Huang (QIYUAN LAB)
RetrievalLarge Language ModelSupervised Fine-TuningReinforcement LearningVision Language ModelMultimodalityRetrieval-Augmented GenerationChain-of-Thought
🎯 What it does: Constructed a multi-modal RAG dataset named VaccineRAG containing chain-of-thought annotations, and trained models to perform self-filtering under retrieval sample diversity.
VAEVQ: Enhancing Discrete Visual Tokenization Through Variational Modeling
Sicheng Yang (Houmo AI), Dawei Yang (Houmo AI)
GenerationTransformerDiffusion modelAuto EncoderImageBiomedical Data
🎯 What it does: Propose the VAEVQ framework, which applies VAE to vector quantization and incorporates RCS and DCR to enhance codebook utilization and generation quality.
VAGU & GtS: LLM-Based Benchmark and Framework for Joint Video Anomaly Grounding and Understanding
Shibo Gao (Beijing Jiaotong University), Linlin Huang (Beijing Jiaotong University)
Anomaly DetectionLarge Language ModelPrompt EngineeringVision Language ModelVideoTextBenchmarkRetrieval-Augmented Generation
🎯 What it does: This paper proposes the VAGU benchmark and the GtS training-free framework to achieve temporal localization and semantic understanding of video anomalies, and introduces the JeAUG metric for unified evaluation.
VALIANT: Prompt Instability for Active Learning in Black-Box Medical Imaging
Dwarikanath Mahapatra, Mauricio Reyes (MBZUAI)
TransformerPrompt EngineeringBiomedical Data
🎯 What it does: To adapt black-box medical imaging foundation models, the VALIANT framework is proposed to achieve active learning;
Variance Computation for Weighted Model Counting with Knowledge Compilation Approach
Kengo Nakamura (NTT, Inc.), Norihito Yasuda (NTT, Inc.)
Computational EfficiencyRepresentation LearningGraph
🎯 What it does: This paper proposes a polynomial-time algorithm for computing the variance of weighted model counting (WMC) results under the structured d-DNNF (st-d-DNNF) representation, and proves that the problem is NP-hard under looser representations (e.g., d-DNNF, FBDD). Subsequently, the algorithm is applied to Bayesian networks with constant tree width, enabling polynomial-time computation of the variance of marginal probabilities, with experiments demonstrating its efficiency on real-world networks.
Variance Reduction via Resampling and Experience Replay
Jiale Han (UCLA), Yuhua Zhu (UCLA)
Reinforcement LearningTabular
🎯 What it does: This paper models experience replay as resampling U- and V-statistics, providing a theoretical framework and proving its variance reduction and stability improvement in policy evaluation (LSTD, PDE-based) and supervised learning (kernel ridge regression).
Variation-Bounded Loss for Noise-Tolerant Learning
Jialiang Wang (Harbin Institute of Technology), Haoliang Li (Harbin Institute of Technology)
ClassificationConvolutional Neural NetworkImage
🎯 What it does: This paper proposes the Variation Ratio, a novel attribute to measure the robustness of loss functions, and defines the Variation-Bounded Loss (VBL) based on this. Through theoretical analysis of the variation ratio, its advantages in noise tolerance are demonstrated. Multiple classical losses (CE, EL, SL, etc.) are rewritten in variation-bounded forms. Subsequent comparisons with existing mainstream robust losses and semi-supervised learning methods on synthetic and real-world noisy datasets show that VBL ranks among the top three or even achieves the best performance in various noise scenarios, while maintaining strong fitting capabilities on noise-free data.
Variational OOD State Correction for Offline Reinforcement Learning
Ke Jiang (Nanjing University of Aeronautics and Astronautics), Xiaoyang Tan (Nanjing University of Aeronautics and Astronautics)
Reinforcement LearningAuto EncoderSequentialBenchmark
🎯 What it does: Studies the state distribution shift problem in offline reinforcement learning, proposing Density-Aware Safety Perception (DASP) based on a variational framework for out-of-distribution (OOD) state correction.
VasoMIM: Vascular Anatomy-Aware Masked Image Modeling for Vessel Segmentation
De-Xing Huang (State Key Laboratory of Multimodal Artificial Intelligence Systems, Institute of Automation, Chinese Academy of Sciences), Zeng-Guang Hou (State Key Laboratory of Multimodal Artificial Intelligence Systems, Institute of Automation, Chinese Academy of Sciences)
SegmentationTransformerAuto EncoderImageBiomedical Data
🎯 What it does: Propose a VasoMIM framework, an occlusion image modeling approach based on vascular anatomy knowledge, specifically designed for X-ray vascular image segmentation;
VBF++: Variational Bayesian Fusion with Context-Aware Priors and Recommendation-Guided Adversarial Refinement for Multimodal Video Recommendation
Ziyi Cao (Beihang University), Yong Chen (Beijing University of Posts and Telecommunications)
Recommendation SystemMeta LearningAuto EncoderGenerative Adversarial NetworkVideoMultimodality
🎯 What it does: Proposed the VBF++ framework, which addresses the uncertainty in fusion strategies and target mismatch in multi-modal video recommendation by combining variational Bayesian fusion with context-aware priors and recommendation-guided adversarial refinement.