AAAI 2025 Papers — Page 9
AAAI Conference on Artificial Intelligence · 3028 papers
Domain-Level Disentanglement Framework Based on Information Enhancement for Cross-Domain Cold-Start Recommendation
Nian Rong (Beijing Jiaotong University), Liang Wang
Recommendation SystemGraph Neural NetworkAuto EncoderContrastive LearningTabular
🎯 What it does: A D2C2R framework based on domain-level decoupling and information enhancement is proposed to address the cross-domain cold start recommendation problem.
DOMAINEVAL: An Auto-Constructed Benchmark for Multi-Domain Code Generation
Qiming Zhu (Chinese Information Processing Laboratory, Institute of Software, Chinese Academy of Sciences), Shing-Chi Cheung (Hong Kong University of Science and Technology)
GenerationAI Code AssistantLarge Language ModelTextBenchmark
🎯 What it does: This paper presents DOMAINEVAL, a multi-domain code generation benchmark that includes 2,454 problems and 5,892 test cases, covering six major domains: computation, networking, basic operations, systems, visualization, and cryptography.
DOMBA: Double Model Balancing for Access-Controlled Language Models via Minimum-Bounded Aggregation
Tom Segal (Ben-Gurion University of the Negev), Yuval Elovici (Ben-Gurion University of the Negev)
Safty and PrivacyComputational EfficiencyTransformerLarge Language ModelSupervised Fine-TuningText
🎯 What it does: The DOMBA method is proposed, which utilizes two sub-models trained on two different access level datasets, and implements access control during inference using minimum-bounded relative probabilities, balancing security and efficiency.
DomCLP: Domain-wise Contrastive Learning with Prototype Mixup for Unsupervised Domain Generalization
Jin-Seop Lee (Sungkyunkwan University), Jee-Hyong Lee (Sungkyunkwan University)
Domain AdaptationRepresentation LearningConvolutional Neural NetworkContrastive LearningImage
🎯 What it does: Proposes the DomCLP method, which utilizes Domain-wise Contrastive Learning (DCon) and Prototype Mixup (PMix) for unsupervised domain generalization.
Don’t Think It Twice: Exploit Shift Invariance for Efficient Online Streaming Inference of CNNs
Christodoulos Kechris (Ecole Polytechnique Federale de Lausanne), David Atienza (Ecole Polytechnique Federale de Lausanne)
Computational EfficiencyConvolutional Neural NetworkTime SeriesBiomedical DataElectrocardiogram
🎯 What it does: This paper proposes an online streaming inference framework named StreamiNNC, which leverages the translational invariance of CNN convolutions to skip unnecessary computations under overlapping windows, significantly reducing the inference cost of temporal CNNs.
Doubly Contrastive Learning for Source-Free Domain Adaptive Person Search
Yizhen Jia (Nanjing University of Aeronautics and Astronautics), Jie Qin (Nanjing University of Aeronautics and Astronautics)
Object DetectionDomain AdaptationGraph Neural NetworkContrastive LearningImage
🎯 What it does: This paper proposes the Source-Free Domain Adaptive Person Search (SFDA-PS) task and presents a Dual Contrastive Learning (DCL) framework.
Doubly-Bounded Queue for Constrained Online Learning: Keeping Pace with Dynamics of Both Loss and Constraint
Juncheng Wang (Hong Kong Baptist University), Yituo Liu (Hong Kong Baptist University)
OptimizationTabularTime Series
🎯 What it does: The COLDQ algorithm is proposed, which implements dynamic constraint control in online convex optimization using a double-bound virtual queue, balancing dynamic regret and hard constraint violations.
DP-MemArc: Differential Privacy Transfer Learning for Memory Efficient Language Models
Yanming Liu (Zhejiang University), Tianyu Du (Renmin University of China)
Safty and PrivacyComputational EfficiencyTransformerLarge Language ModelSupervised Fine-TuningText
🎯 What it does: The DP-MemArc framework is proposed, which significantly reduces GPU memory and RAM during fine-tuning large language models under differential privacy through the use of a side network (DP-MemArcside) and a reversible network (DP-MemArcrev), while maintaining model accuracy.
DPCL-Diff：Temporal Knowledge Graph Reasoning Based on Graph Node Diffusion Model with Dual-Domain Periodic Contrastive Learning
Yukun Cao (Shanghai University of Electric Power), Luobin Huang (Shanghai University of Electric Power)
Graph Neural NetworkTransformerDiffusion modelContrastive LearningGraphTime Series
🎯 What it does: In the task of temporal knowledge graph reasoning, the DPCL-Diff model is proposed, which combines a graph node diffusion model and dual-domain periodic contrastive learning to achieve high-quality predictions for new and periodic events.
DPLUT: Unsupervised Low-light Image Enhancement with Lookup Tables and Diffusion Priors
Yunlong Lin (Xiamen University), Xinghao Ding (Xiamen University)
RestorationConvolutional Neural NetworkDiffusion modelImage
🎯 What it does: This paper proposes an unsupervised low-light image enhancement framework called DPLUT, which utilizes a two-stage lookup table (LLUT and NLUT) to achieve high-quality and real-time low-light recovery.
DR-VAE: Debiased and Representation-enhanced Variational Autoencoder for Collaborative Recommendation
Fan Wang (Zhejiang University), Jianwei Yin (Zhejiang University)
Recommendation SystemAuto EncoderTabularOrdinary Differential Equation
🎯 What it does: The DR-VAE framework is proposed, combining a debiasing estimator and a continuous representation enhancer to address exposure bias and posterior collapse issues in collaborative filtering.
Dr. Tongue: Sign-Oriented Multi-label Detection for Remote Tongue Diagnosis
Yiliang Chen (Hong Kong Polytechnic University), Jing Qin (Hong Kong Polytechnic University)
RecognitionObject DetectionSegmentationConvolutional Neural NetworkImage
🎯 What it does: This paper proposes a multi-label attribute detection framework for remote tongue diagnosis called SignNet, and constructs a novel tongue image dataset.
Drawing Informative Gradients from Sources: A One-stage Transfer Learning Framework for Cross-city Spatiotemporal Forecasting
Yudong Zhang (University of Science and Technology of China), Yang Wang (University of Science and Technology of China)
Domain AdaptationMeta LearningGraph Neural NetworkReinforcement LearningPrompt EngineeringTime Series
🎯 What it does: A single-stage transfer learning framework for cross-city spatiotemporal prediction, TSJT, is proposed. This framework filters and enhances the source city gradients through the Target-Skewed Backward training strategy and Node Prompting Module, allowing for joint training directly on source and target city data.
DREAM: Decoupled Discriminative Learning with Bigraph-aware Alignment for Semi-supervised 2D-3D Cross-modal Retrieval
Fan Zhang (Georgia Institute of Technology), Xiao Luo (University of California Los Angeles)
RetrievalConvolutional Neural NetworkGraph Neural NetworkContrastive LearningImagePoint Cloud
🎯 What it does: A semi-supervised 2D-3D cross-modal retrieval framework named DREAM is proposed, which separates label prediction from reliability assessment and utilizes a large graph to achieve modality alignment.
DreamAlign: Dynamic Text-to-3D Optimization with Human Preference Alignment
Gaofeng Liu (Shanghai Jiao Tong University), Tao Fang (Shanghai Jiao Tong University)
GenerationOptimizationDiffusion modelContrastive LearningPoint Cloud
🎯 What it does: Developed the DreamAlign system, achieving direct alignment from text to 3D generation based on human preferences.
DreaMark: Rooting Watermark in Score Distillation Sampling Generated Neural Radiance Fields
Xingyu Zhu (Southern University of Science and Technology), Xuetao Wei (Hong Kong Polytechnic University)
GenerationData SynthesisDiffusion modelScore-based ModelNeural Radiance FieldImage
🎯 What it does: During the NeRF generation process, a watermark is embedded through Score Distillation Sampling (SDS), rooting key information in the generative model;
DreamFit: Garment-Centric Human Generation via a Lightweight Anything-Dressing Encoder
Ente Lin (Shenzhen International Graduate School Tsinghua University), Xiaodan Liang (Sun Yat-sen University)
GenerationData SynthesisLarge Language ModelDiffusion modelImageMultimodality
🎯 What it does: This paper presents DreamFit, a lightweight clothing-centered portrait generation framework that constructs an Anything-Dressing Encoder using LoRA layers and incorporates reference image features into the pre-trained Stable Diffusion UNet through adaptive attention, generating high-quality clothing detail images.
DreamPhysics: Learning Physics-Based 3D Dynamics with Video Diffusion Priors
Tianyu Huang (Harbin Institute of Technology), Rynson W. H. Lau (City University of Hong Kong)
GenerationData SynthesisKnowledge DistillationDiffusion modelScore-based ModelVideoPhysics Related
🎯 What it does: Utilizing the physical priors of video diffusion models to learn the material properties of 3D Gaussian point fields, and then simulating dynamics through Material Point Method (MPM) to ultimately generate 4D content that complies with physical laws.
DreamUHD: Frequency Enhanced Variational Autoencoder for Ultra-High-Definition Image Restoration
Yidi Liu (University of Science and Technology of China), Xueyang Fu (University of Science and Technology of China)
RestorationSuper ResolutionAuto EncoderImage
🎯 What it does: A super high-resolution image restoration framework based on frequency domain priors, called Frequency Domain Variational Autoencoder (FE-VAE) with a wavelet adapter, is designed to perform various UHD image restoration tasks in a low-parameter and efficient latent space.
DriveDreamer-2: LLM-Enhanced World Models for Diverse Driving Video Generation
Guosheng Zhao (University of Chinese Academy of Sciences), Xingang Wang
GenerationData SynthesisAutonomous DrivingTransformerLarge Language ModelSupervised Fine-TuningDiffusion modelWorld ModelVideoText
🎯 What it does: This paper presents DriveDreamer-2, a world model based on LLM that generates multi-view driving videos from text descriptions, supporting customizable scenarios such as vehicle intrusions.
DriveEditor: A Unified 3D Information-Guided Framework for Controllable Object Editing in Driving Scenes
Yiyuan Liang (Huazhong University of Science and Technology), Xu Zou (Huazhong University of Science and Technology)
GenerationAutonomous DrivingDiffusion modelVideo
🎯 What it does: We propose DriveEditor, a unified framework based on diffusion models that enables object relocation, insertion, replacement, and deletion in driving scene videos.
DriveGazen: Event-Based Driving Status Recognition Using Conventional Camera
Xiaoyin Yang (Dalian University of Technology), Xin Yang (Dalian University of Technology)
RecognitionAutonomous DrivingSpiking Neural NetworkVideo
🎯 What it does: Developed a wearable eye camera and a driving state recognition system based on event frames generated by traditional cameras.
Driving in the Occupancy World: Vision-Centric 4D Occupancy Forecasting and Planning via World Models for Autonomous Driving
Yu Yang (Zhejiang University), Yong Liu (Zhejiang University)
Autonomous DrivingTransformerWorld ModelPoint Cloud
🎯 What it does: We propose Drive-OccWorld, a vision-based 4D occupancy prediction and planning world model that can generate controllable future occupancy states based on different driving action conditions and achieve end-to-end safe trajectory planning on this basis.
DrivingForward: Feed-forward 3D Gaussian Splatting for Driving Scene Reconstruction from Flexible Surround-view Input
Qijian Tian (Shanghai Jiao Tong University), Lizhuang Ma (East China Normal University)
Pose EstimationDepth EstimationAutonomous DrivingOptimizationConvolutional Neural NetworkGaussian SplattingImage
🎯 What it does: We propose DrivingForward, a feedforward model based on 3D Gaussian splatting that can reconstruct driving scenes in real-time and supports multi-frame input with any number of surrounding views.
Drop the Beat! Freestyler for Accompaniment Conditioned Rapping Voice Generation
Ziqian Ning (Northwestern Polytechnical University), Lei Xie (Microsoft)
GenerationTransformerLarge Language ModelAudio
🎯 What it does: Freestyler has been developed - the first model capable of directly generating rap vocals based on lyrics and accompaniment, and a large-scale rap dataset called RapBank has been constructed.
DrugHash: Hashing Based Contrastive Learning for Virtual Screening
Jin Han (Nanjing University), Wu-Jun Li (Nanjing University)
Drug DiscoveryTransformerContrastive LearningBiomedical Data
🎯 What it does: A hash-based contrastive learning framework called DrugHash is proposed, which uses binary hash codes to achieve efficient virtual screening.
DSGram: Dynamic Weighting Sub-Metrics for Grammatical Error Correction in the Era of Large Language Models
Jinxiang Xie (Beijing Jiaotong University), Xiaojun Wan (Peking University)
Large Language ModelPrompt EngineeringText
🎯 What it does: A framework named DSGram is proposed for the dynamic weighted sub-indicator evaluation of grammar error correction (GEC) models, implementing sub-indicator scoring and weight generation based on LLM.
DSRC: Learning Density-Insensitive and Semantic-Aware Collaborative Representation Against Corruptions
Jingyu Zhang (Fudan University), Liang Song (Fudan University)
Autonomous DrivingKnowledge DistillationRepresentation LearningPoint Cloud
🎯 What it does: A DSRC framework is proposed, which enhances the robustness of multi-agent collaborative perception under natural disturbances through sparse-to-dense knowledge distillation and semantically guided point cloud re-rendering.
Dual Conditioned Motion Diffusion for Pose-Based Video Anomaly Detection
Hongsong Wang (Southeast University), Jie Gui (Southeast University)
Anomaly DetectionTransformerDiffusion modelAuto EncoderVideo
🎯 What it does: A dual conditional motion diffusion framework (DCMD) that integrates reconstruction and prediction is proposed for video anomaly detection based on skeletal poses.
Dual Information Purification for Lightweight SAR Object Detection
Xi Yang (Xidian University), De Cheng (Xidian University)
Object DetectionKnowledge DistillationImage
🎯 What it does: The Dual Information Purification Knowledge Distillation (DIPKD) method is proposed for lightweight SAR target detection knowledge distillation, which includes three main modules: Selective Noise Suppression (SNS), Knowledge Layer Separation (KLD), and Reverse Information Transmission (RIT);
Dual Manifold Regularization Steered Robust Representation Learning for Point Cloud Analysis
Jian Bi (Nanjing University of Science and Technology), Jian Yang (Nanjing University of Science and Technology)
Representation LearningContrastive LearningPoint Cloud
🎯 What it does: A dual hyperbolic space and spherical space dual hyperbolic regularization (DMR) framework is proposed for point cloud representation learning, enhancing robustness through regularization.
Dual-branch Graph Feature Learning for NLOS Imaging
Xiongfei Su (Zhejiang University), Xin Yuan (University of Science and Technology of China)
RestorationDepth EstimationGraph Neural NetworkImageGraph
🎯 What it does: This paper proposes a dual-branch graph neural network framework (DG-NLOS) for simultaneously recovering high albedo and depth information in non-line-of-sight (NLOS) scenarios.
Dual-calibrated Co-training Framework for Personalized Federated Semi-Supervised Medical Image Segmentation
Delin Pan (Jiangnan University), Xiang Pan (Jiangnan University)
SegmentationFederated LearningImageBiomedical Data
🎯 What it does: A personalized federated semi-supervised learning framework is proposed for medical image segmentation tasks, utilizing dual calibration co-training to enhance model performance.
Dual-Channel Interactive Graph Transformer for Traffic Classification with Message-Aware Flow Representation
Xing Qiu (Southeast University), Nan Fu (Southeast University)
ClassificationGraph Neural NetworkTransformerGraph
🎯 What it does: A dual-channel interactive graph Transformer (DigTraffic) is proposed for traffic classification, utilizing a message-level interactive graph (MTIG) to construct packet-level nodes and designing three types of heterogeneous edges. It combines dual-channel encoding of packet length and timing, and incorporates centrality, spatial, and edge encoding in the Transformer to capture global structural information.
Dual-Level Precision Edges Guided Multi-View Stereo with Accurate Planarization
Kehua Chen (Institute of Computing Technology, Chinese Academy of Sciences), Zhaoqi Wang (Institute of Computing Technology, Chinese Academy of Sciences)
Depth EstimationOptimizationImage
🎯 What it does: In multi-view stereo reconstruction, a method based on dual-layer fine edges (DPE-MVS) is proposed to address the issue of constructing plane models in low-texture and randomly textured areas. This method uses fine edges and coarse edges to jointly guide plane construction and matching.
Dual-View Learning for Conversational Emotion Recognition Through Context and Emotion-Shift Modeling
Xupeng Zha (Hunan University), Zixing Zhang (Hunan University)
RecognitionRecurrent Neural NetworkGraph Neural NetworkTransformerText
🎯 What it does: A dual-view learning framework DVL-CER is proposed, which captures the speaker's emotional transition through an emotional view and shares the emotional distribution with the dialogue context view.
DualCP: Rehearsal-Free Domain-Incremental Learning via Dual-Level Concept Prototype
Qiang Wang (Xi'an Jiaotong University), Yihong Gong (Shenzhen University of Advanced Technology)
ClassificationDomain AdaptationTransformerContrastive LearningImage
🎯 What it does: This paper proposes a replay-free domain incremental learning method called DualCP, which constructs dual-layer concept prototypes (coarse-grained and fine-grained) to achieve a unified representation of similar categories across domains, thereby addressing the conflict between learning new domains and retaining old domain memories.
DualDynamics: Synergizing Implicit and Explicit Methods for Robust Irregular Time Series Analysis
YongKyung Oh (University of California), Sungil Kim (Ulsan National Institute of Science and Technology)
ClassificationAnomaly DetectionRecurrent Neural NetworkFlow-based ModelTime SeriesSequentialBiomedical DataBenchmarkStochastic Differential EquationOrdinary Differential Equation
🎯 What it does: The DualDynamics framework is proposed, which integrates implicit models based on NDE with explicit models of reversible neural flows to handle irregular and missing time series.
DualGFL: Federated Learning with a Dual-Level Coalition-Auction Game
Xiaobing Chen (Louisiana State University), Mingxuan Sun (Louisiana State University)
Federated LearningImage
🎯 What it does: Proposes the DualGFL framework, which integrates lower-level cooperative alliance games and upper-level multi-attribute auction games to achieve client collaboration and competitive decision-making in federated learning.
DualNet: Robust Self-Supervised Stereo Matching with Pseudo-Label Supervision
Yun Wang (City University of Hong Kong), Junjie Hu (Shenzhen Institute of Artificial Intelligence and Robotics for Society)
Depth EstimationDomain AdaptationKnowledge DistillationContrastive LearningImage
🎯 What it does: Proposes the DualNet framework, which first trains a self-supervised teacher model and then uses high-quality pseudo-labels to train a student model, achieving self-supervised stereo matching.
DualOpt: A Dual Divide-and-Optimize Algorithm for the Large-scale Traveling Salesman Problem
Shipei Zhou (Huazhong University of Science and Technology), Yan Jin (Huazhong University of Science and Technology)
OptimizationTabular
🎯 What it does: This paper proposes DualOpt, a dual divide-and-conquer optimization algorithm for solving large-scale traveling salesman problems.
DuMo: Dual Encoder Modulation Network for Precise Concept Erasure
Feng Han (Fudan University), Yu-Gang Jiang (Fudan University)
GenerationData SynthesisDiffusion modelImageText
🎯 What it does: In the text-to-image diffusion model, a module is proposed that only modifies the skip connection features, achieving precise elimination of target concepts while maintaining the generation quality of non-target concepts.
Dung’s Argumentation Framework: Unveiling the Expressive Power with Inconsistent Databases
Yasir Mahmood (Paderborn University), Axel-Cyrille Ngonga Ngomo (Paderborn University)
🎯 What it does: It is proven that Dung's abstract argumentation framework can be viewed as an inconsistent database using only functional dependencies (FD) and inclusion dependencies (ID), and an equivalent mapping between the two is established.
DUO: Diverse, Uncertain, On-Policy Query Generation and Selection for Reinforcement Learning from Human Feedback
Xuening Feng (Shanghai Jiao Tong University), Yifei Zhu (Shanghai Jiao Tong University)
Reinforcement Learning from Human FeedbackReinforcement LearningSequential
🎯 What it does: The DUO method is proposed, designed for the human feedback (RLHF) task in reinforcement learning, featuring a diverse, confidence-uncertain, and policy-relevant query generation and selection process that significantly improves query efficiency and enhances the final policy performance.
DuSSS: Dual Semantic Similarity-Supervised Vision-Language Model for Semi-Supervised Medical Image Segmentation
Qingtao Pan (Shandong University), Shuo Li (Case Western Reserve University)
SegmentationTransformerVision Language ModelContrastive LearningImageBiomedical DataComputed Tomography
🎯 What it does: Combining visual-language models with semi-supervised medical image segmentation, the DuSSS method is proposed, which utilizes dual contrastive learning and semantic similarity supervision to enhance cross-modal consistency. It also improves the quality of pseudo-labels through text-guided pseudo-label generation and a teacher-student framework, ultimately achieving more accurate semi-supervised segmentation.
DUSTED: Dual-Attention Enhanced Spatial Transcriptomics Denoiser
Jun Zhu (Tsinghua University), Cheng Chang (National Center for Protein Sciences)
RestorationGraph Neural NetworkAuto EncoderBiomedical Data
🎯 What it does: A denoising method for spatial transcriptomics (SRT) called DUSTED is proposed, which can restore high-quality spatial transcriptomic data without using external images.
DutyTTE: Deciphering Uncertainty in Origin-Destination Travel Time Estimation
Xiaowei Mao (Beijing Jiaotong University), Huaiyu Wan (Beijing Jiaotong University)
OptimizationTransformerReinforcement LearningMixture of ExpertsTime SeriesSequential
🎯 What it does: Developed the DutyTTE method, which first optimizes OD path prediction using deep reinforcement learning, and then quantifies segment-level travel time uncertainty through Mixture of Experts.
DVP-MVS: Synergize Depth-Edge and Visibility Prior for Multi-View Stereo
Zhenlong Yuan (Institute of Computing Technology), Zhaoqi Wang (Institute of Computing Technology)
RestorationDepth EstimationImageBenchmark
🎯 What it does: This paper proposes DVP-MVS, a robust and visually perceptive patch deformation multi-view stereo reconstruction method achieved through depth-edge alignment priors and cross-view priors.
dyAb: Flow Matching for Flexible Antibody Design with AlphaFold-driven Pre-binding Antigen
Cheng Tan (Tencent), Stan Z. Li (Westlake University)
Drug DiscoveryProtein Structure PredictionBiomedical DataOrdinary Differential Equation
🎯 What it does: This paper proposes the dyAb framework, which utilizes the pre-bound antigen structures predicted by AlphaFold2 and completes the antibody design and binding process modeling through coarse-grained interface alignment and fine-grained flow matching.
Dynamic Adapter with Semantics Disentangling for Cross-lingual Cross-modal Retrieval
Rui Cai (Zhejiang Gongshang University), Xun Wang (Zhejiang Gongshang University)
RetrievalDomain AdaptationTransformerContrastive LearningImageVideoTextMultimodality
🎯 What it does: This paper proposes a dynamic adapter and semantic decoupling method for cross-language and cross-modal retrieval, aiming to achieve alignment between visual data and low-resource languages without the need for labeled data in the target language.
Dynamic Clustering Convolutional Neural Network
Tanzhe Li (Xiamen University), Taisong Jin (Anyang Normal University)
ClassificationObject DetectionSegmentationConvolutional Neural NetworkImage
🎯 What it does: This paper proposes a convolutional network DCCNeXt based on global clustering, which utilizes dynamic clustering convolution (DCConv) to group image patches into semantically similar clusters and extracts features using shared convolutional kernels.
Dynamic Contrastive Knowledge Distillation for Efficient Image Restoration
Yunshuai Zhou (East China Normal University), Shaohui Lin (Huawei Noah's Ark Lab)
RestorationSuper ResolutionKnowledge DistillationContrastive LearningImage
🎯 What it does: A dynamic contrastive knowledge distillation framework (DCKD) is proposed for compressing image restoration models, mainly consisting of dynamic contrastive regularization and distribution mapping modules.
Dynamic Entity-Masked Graph Diffusion Model for Histopathology Image Representation Learning
Zhenfeng Zhuang (Xiamen University), Liansheng Wang (Xiamen University)
ClassificationRepresentation LearningGraph Neural NetworkDiffusion modelAuto EncoderImageBiomedical Data
🎯 What it does: A self-supervised pathological image representation learning method based on a dynamic entity masking graph diffusion model (H-MGDM) is proposed, which captures the topological relationships of tissue entities using graph structures and reconstructs masked subgraphs through graph diffusion.
Dynamic Expansion Diffusion Learning for Lifelong Generative Modelling
Fei Ye (University of Electronic Science and Technology of China), Kun Zhang (Carnegie Mellon University)
GenerationData SynthesisKnowledge DistillationRepresentation LearningDiffusion modelAuto EncoderImage
🎯 What it does: A Dynamic Expansion Diffusion Model (DEDM) has been designed and implemented, capable of continuously adding new diffusion components in an online continual learning scenario without task boundaries, while retaining memory of learned knowledge without catastrophic forgetting.
Dynamic Graph Learning with Static Relations for Credit Risk Assessment
Qi Yuan (University of Chinese Academy of Sciences), Xiang Ao (University of Chinese Academy of Sciences)
ClassificationRecommendation SystemAnomaly DetectionOptimizationExplainability and InterpretabilityComputational EfficiencyGraph Neural NetworkGraphTabularFinance Related
🎯 What it does: This paper proposes a credit risk assessment framework that combines dynamic graph neural networks with static relationships (DGNN-SR), utilizing a user-merchant dynamic transaction graph and a user-user static transfer graph for joint modeling.
Dynamic Interactive Bimodal Hypergraph Networks for Emotion Recognition in Conversations
Xuping Chen (Shenzhen University), Wuzhen Shi (Shenzhen University)
RecognitionRecurrent Neural NetworkGraph Neural NetworkMultimodality
🎯 What it does: A dynamic interactive bimodal hypergraph convolutional network (DIB-HGCN) is proposed for emotion recognition in conversations (ERC).
Dynamic Multi-Interest Graph Neural Network for Session-Based Recommendation
Mingyang Lv (Jilin University), Yuanbo Xu (Jilin University)
Recommendation SystemGraph Neural NetworkGraphSequential
🎯 What it does: A Dynamic Multi-Interest Graph Neural Network (DMI-GNN) is proposed for conversational recommendation, modeling conversational data as a graph and extracting multi-interest representations.
Dynamic Neighborhood Modeling via Node-Subgraph Contrastive Learning for Graph-Based Fraud Detection
Zhizhi Yu (Tianjin University), Jianguo Wei (Tianjin University)
Anomaly DetectionGraph Neural NetworkContrastive LearningGraphFinance Related
🎯 What it does: A dynamic neighborhood modeling method DCL-GFD based on node-subgraph contrastive learning is proposed for graph fraud detection.
Dynamic Operator Optimization for Efficient Multi-Tenant LoRA Model Serving
Changhai Zhou (Fudan University), Zekai Liu (Fudan University)
OptimizationComputational EfficiencyTransformerLarge Language ModelText
🎯 What it does: A dynamic operator optimization framework Dop is built for multi-tenant LoRA model services, specifically implementing adaptive and performance-optimal CUDA code for SGMV operations.
Dynamic Spectral Graph Anomaly Detection
Jianbo Zheng (Hunan University), Xianxun Zhu (Shanghai University)
Anomaly DetectionGraph Neural NetworkGraphFinance Related
🎯 What it does: This paper proposes a dynamic spectral graph anomaly detection framework called DSGAD, which improves upon traditional manually designed wave functions and feature concatenation methods by using learnable Beta-mixed wave functions and channel convolution fusion.
Dynamic Syntactic Feature Filtering and Injecting Networks for Cross-lingual Dependency Parsing
Jianjian Liu (Kunming University of Science and Technology), Shengxiang Gao (Kunming University of Science and Technology)
Recurrent Neural NetworkTransformerLarge Language ModelText
🎯 What it does: This paper proposes a dynamic syntactic feature filtering and injection network based on a shared-private framework for cross-lingual dependency parsing.
Dynamic Target Distribution Estimation for Source-Free Open-Set Domain Adaptation
Zhiqi Yu (University of Electronic Science and Technology of China), Lei Zhu (University of Queensland)
Domain AdaptationImage
🎯 What it does: A source-independent open set domain adaptation method called DTDE is proposed, which dynamically constructs known and unknown class prototypes based on the clustering distribution of the target domain, utilizing self-supervised learning and class-aware KNN alignment for model adaptation.
Dynamic Uncertainty Estimation for Offline Reinforcement Learning
Jiesheng Wang (Shanxi University), Xin Yang (Southwestern University of Finance and Economics)
Reinforcement LearningGenerative Adversarial NetworkTabular
🎯 What it does: This paper proposes an offline reinforcement learning algorithm called DURL, which is based on dynamic uncertainty estimation and density truncation for OOD data sampling, enabling better Q-value estimation in situations where the environment is not fully known.
Dynamic-Width Speculative Beam Decoding for LLM Inference
Zongyue Qin (University of California Los Angeles), Yizhou Sun (University of California Los Angeles)
OptimizationComputational EfficiencyTransformerLarge Language ModelText
🎯 What it does: A dynamic width speculative beam search (DSBD) algorithm is proposed, which combines speculative decoding with beam search to improve the inference efficiency and output quality of large language models.
DynASyn: Multi-Subject Personalization Enabling Dynamic Action Synthesis
Yongjin Choi (Korea University), Seung Jun Baek (Korea University)
GenerationData SynthesisPose EstimationVision Language ModelDiffusion modelImageMultimodality
🎯 What it does: This paper proposes DynASyn, a technique capable of personalized generation from a single multi-subject image, which can generate diverse poses, actions, and interactions while maintaining the identity of the subjects.
DZAD: Diffusion-based Zero-shot Anomaly Detection
Tianrui Zhang (National University of Singapore), Yiping Gao (Huazhong University of Science and Technology)
Anomaly DetectionDiffusion modelImage
🎯 What it does: A zero-shot anomaly detection framework DZAD based on diffusion models is proposed, which achieves anomaly localization through multi-temporal noise feature extraction and a semantic guidance network without the need for text prompts.
E4: Energy-Efficient DNN Inference for Edge Video Analytics via Early Exiting and DVFS
Ziyang Zhang (Harbin Institute of Technology), Jie Liu (Harbin Institute of Technology)
RecognitionOptimizationComputational EfficiencyRecurrent Neural NetworkVideo
🎯 What it does: The E4 framework is proposed, which combines early exit based on video frame complexity with dynamic frequency scaling to enhance the energy efficiency of edge video analysis.
Each Fake News Is Fake in Its Own Way: An Attribution Multi-Granularity Benchmark for Multimodal Fake News Detection
Hao Guo (Nation University of Defense Technology), Xiang Zhao (Nation University of Defense Technology)
ClassificationTransformerLarge Language ModelVision Language ModelImageTextMultimodalityBenchmark
🎯 What it does: This paper proposes the task of attribution and detection of multimodal fake news, constructs the first fine-grained attribution dataset AMG, and designs a multi-granularity clue alignment model MGCA.
Early Concept Drift Detection via Prediction Uncertainty
Pengqian Lu (Australian Artificial Intelligence Institute), Guangquan Zhang (Australian Artificial Intelligence Institute)
Anomaly DetectionConvolutional Neural NetworkImageTabular
🎯 What it does: This paper proposes the Prediction Uncertainty Index (PU-index) and its concept drift detector based on Chi-square test, PUDD, and implements online drift detection through an adaptive binning algorithm.
EBBS: An Ensemble with Bi-Level Beam Search for Zero-Shot Machine Translation
Yuqiao Wen (University of Alberta), Lili Mou (University of Alberta)
Computational EfficiencyKnowledge DistillationTransformerText
🎯 What it does: The EBBS (Ensemble with Bi-Level Beam Search) method is proposed to improve translation quality in multilingual zero-shot machine translation by combining direct translation with various pivot paths, and further utilizing the high-quality translations generated by EBBS for knowledge distillation to enhance model inference efficiency.
EBS-CFL: Efficient and Byzantine-robust Secure Clustered Federated Learning
Zhiqiang Li (East China Normal University), Hong-Ning Dai
Federated LearningSafty and PrivacyImage
🎯 What it does: Designed and implemented an efficient, Byzantine robust secure aggregation framework EBS-CFL to protect user clustering identity and gradient privacy in cluster federated learning.
EchoDiffusion: Waveform Conditioned Diffusion Models for Echo-Based Depth Estimation
Wenjie Zhang (Zhengzhou University), Mingliang Xu (Zhengzhou University)
Depth EstimationDiffusion modelImageAudio
🎯 What it does: The EchoDiffusion framework is proposed, which encodes the acoustic fingerprint spectrum into a latent space and uses sound waveforms to guide the diffusion process to generate depth maps.
EchoMimic: Lifelike Audio-Driven Portrait Animations through Editable Landmark Conditions
Zhiyuan Chen (Ant Group), Chenguang Ma (Ant Group)
GenerationData SynthesisDiffusion modelVideoMultimodalityAudio
🎯 What it does: This paper proposes the EchoMimic framework, which can generate realistic avatar videos using audio, facial keypoints, or a combination of both.
Eco Search: A No-delay Best-First Search Algorithm for Program Synthesis
Théo Matricon (Laboratory of Computer Science and Systems), Guillaume Lagarde (Laboratory of Computer Science and Systems)
Data SynthesisOptimizationTabular
🎯 What it does: A new optimal priority push search algorithm called ECO SEARCH is proposed, which can achieve constant delay enumeration in program synthesis.
EDENet: Echo Direction Encoding Network for Place Recognition Based on Ground Penetrating Radar
Pengyu Zhang (National University of Defense Technology), Liang Shen (National University of Defense Technology)
RecognitionRetrievalConvolutional Neural NetworkTime Series
🎯 What it does: Proposes the EDENet network, which utilizes learnable Gabor filters and direction-aware attention to geometrically encode ground-penetrating radar (GPR) echo sequences, generating compact local recognition descriptors.
Edge Contrastive Learning: An Augmentation-Free Graph Contrastive Learning Model
Yujun Li (Northwestern Polytechnical University), Yuan Yuan (Northwestern Polytechnical University)
ClassificationRepresentation LearningGraph Neural NetworkContrastive LearningGraph
🎯 What it does: Without using any graph data augmentation, the AFECL (Augmentation-Free Edge Contrastive Learning) model is proposed, which generates edge features using node embeddings and performs contrastive learning at the edge level.
EDGE: Unknown-aware Multi-label Learning by Energy Distribution Gap Expansion
Yuchen Sun (Institute of Computing Technology, Chinese Academy of Sciences), Junwei He (Institute of Computing Technology, Chinese Academy of Sciences)
ClassificationAnomaly DetectionConvolutional Neural NetworkSupervised Fine-TuningImage
🎯 What it does: This paper proposes the EDGE framework, which combines multi-label OOD detection with imbalanced learning, utilizing self-supervised energy distribution difference expansion to enhance the OOD discrimination capability of tail samples.
EditBoard: Towards a Comprehensive Evaluation Benchmark for Text-Based Video Editing Models
Yupeng Chen (Chinese University of Hong Kong), Qian Xie (University of Leeds)
GenerationData SynthesisOptical FlowVideoTextBenchmark
🎯 What it does: EditBoard is proposed, providing a comprehensive evaluation benchmark that includes nine automatic metrics and four assessment dimensions (fidelity, execution, consistency, style) for the systematic evaluation of text-driven video editing models.
Editing Memories Through Few Targeted Neurons
Wei Zhou (Huazhong University of Science and Technology), Fei Wang (Ping An Property and Casualty Insurance Company of China)
TransformerLarge Language ModelSupervised Fine-TuningText
🎯 What it does: This paper achieves local editing of factual knowledge in large language models by identifying only about 1% of high-contribution neurons in the model, fine-tuning them, and performing data augmentation with samples that share the same knowledge relationships.
EF2X Exists for Four Agents
Arash Ashuri (Sharif University of Technology), Alkmini Sgouritsa (Athens University of Economics and Business)
Optimization
🎯 What it does: The paper proves the existence of an EF2X allocation (i.e., after any two items are removed, no agent has a preference for the bundles of other agents) in the case of four agents and an arbitrary number of items with cancelable valuations, and provides a pseudo-polynomial time constructive algorithm; it also offers a polynomial time algorithm for the case of three agents.
Effective and Efficient Representation Learning for Flight Trajectories
Shuo Liu (University of Chinese Academy of Sciences), Jingping Bi (Chinese Academy of Sciences)
Anomaly DetectionRepresentation LearningTransformerTime Series
🎯 What it does: The FLIGHT2VEC framework is proposed for unified representation learning of flight trajectories, addressing issues of uneven behavior density and 3D spatial continuity.
Effective Diffusion Transformer Architecture for Image Super-Resolution
Kun Cheng (Xidian University), Jie Hu (Chongqing University of Posts and Telecommunications)
RestorationSuper ResolutionTransformerDiffusion modelImage
🎯 What it does: This paper proposes a diffusion Transformer architecture called DiT-SR, which is trained from scratch. It utilizes a U-shaped full transformer to achieve multi-scale feature extraction and enhances super-resolution quality through frequency-adaptive time step conditioning.
Effects of Momentum in Implicit Bias of Gradient Flow for Diagonal Linear Networks
Bochen Lyu (DataCanvas), Zhanxing Zhu (University of Leeds)
OptimizationTabularOrdinary Differential Equation
🎯 What it does: The research investigates the implicit bias of momentum methods (Heavy-Ball and Nesterov Accelerated Gradient) in diagonal linear networks, revealing their differences from gradient descent;
Efficient 3D Recognition with Event-driven Spike Sparse Convolution
Xuerui Qiu (Institute of Automation, Chinese Academy of Sciences), Guoqi Li (Institute of Automation, Chinese Academy of Sciences)
ClassificationObject DetectionSegmentationAutonomous DrivingComputational EfficiencySpiking Neural NetworkPoint Cloud
🎯 What it does: This paper proposes an event-driven Sparse Spiking Convolution (SSC) and Spike Voxel Coding (SVC) aimed at sparse 3D point clouds, integrating them into the E-3DSNN backbone network to efficiently handle 3D classification, detection, and segmentation tasks.
Efficient Anomaly Detection of Irregular Sequences in Ct-Echo Model Space
Ao Chen (University of Science and Technology of China), Huanhuan Chen (University of Science and Technology of China)
Anomaly DetectionTime SeriesSequentialOrdinary Differential Equation
🎯 What it does: A model space based on Continuous Time Echo Networks (Ct-Echo) is proposed to detect anomalies in irregularly sampled sequences.
Efficient Attention-Sharing Information Distillation Transformer for Lightweight Single Image Super-Resolution
Karam Park (Seoul National University), Nam Ik Cho (Seoul National University)
Super ResolutionComputational EfficiencyKnowledge DistillationTransformerImage
🎯 What it does: A lightweight Transformer framework ASID is proposed, achieving single image super-resolution through information distillation structure, attention sharing, and channel splitting.
Efficient Communication in Multi-Agent Reinforcement Learning with Implicit Consensus Generation
Dapeng Li (Chinese Academy of Sciences), Guoliang Fan (Chinese Academy of Sciences)
Reinforcement LearningAuto EncoderSequential
🎯 What it does: This paper proposes a multi-agent learning framework named COCOM, which combines explicit communication and implicit consensus learning. It generates consensus based on local observations during execution and creates messages to achieve cooperation.
Efficient Connectivity-Preserving Instance Segmentation with Supervoxel-Based Loss Function
Anna Grim (Allen Institute), Uygar Sümbül
SegmentationComputational EfficiencyImageBiomedical Data
🎯 What it does: This paper proposes a topology-aware loss function based on supervoxels to efficiently maintain the connectivity of neuron instance segmentation.
Efficient Deformable Convolutional Prompt for Continual Test-Time Adaptation in Medical Image Segmentation
Shiyu Liu (Hebei University of Technology), Xiaoke Hao (Hebei University of Technology)
SegmentationDomain AdaptationConvolutional Neural NetworkPrompt EngineeringBiomedical DataMagnetic Resonance Imaging
🎯 What it does: An Efficient Deformable Convolution Prompt (EDCP) is proposed, achieving continuous testing adaptation in medical image segmentation while avoiding large-scale parameter updates on the source model.
Efficient Event-Based Semantic Segmentation via Exploiting Frame-Event Fusion: A Hybrid Neural Network Approach
Hebei Li (University of Science and Technology of China), Xiaoyan Sun (University of Science and Technology of China)
SegmentationAutonomous DrivingConvolutional Neural NetworkSpiking Neural NetworkImageMultimodality
🎯 What it does: A hybrid ANN-SNN framework is proposed to fuse the sparse asynchronous data from event cameras with the spatial information of frame images, achieving efficient semantic segmentation.
Efficient Fault-Tolerant Search by Fast Indexing of Subnetworks
Davide Bilò (University of L'Aquila), Martin Schirneck (University of Vienna)
OptimizationComputational EfficiencyGraph Neural NetworkGraph
🎯 What it does: This study proposes a technique for a fast index subnetwork to construct an efficient fault-tolerant search structure, mainly targeting network problems such as L-hop shortest paths, k-paths, and k-cliques;
Efficient Few-Shot Neural Architecture Search by Counting the Number of Nonlinear Functions
Youngmin Oh (Yonsei University), Bumsub Ham (Korea Institute of Science and Technology)
Neural Architecture SearchImage
🎯 What it does: Proposes a Few-Shot NAS method based on the counting of nonlinear functions in sub-networks.
Efficient Gaussian Splatting for Monocular Dynamic Scene Rendering via Sparse Time-Variant Attribute Modeling
Hanyang Kong (National University of Singapore), Xinchao Wang (National University of Singapore)
Data SynthesisComputational EfficiencyNeural Radiance FieldGaussian SplattingPoint Cloud
🎯 What it does: This paper proposes an Efficient Dynamic Gaussian Splatting (EDGS) based on a sparse anchor-grid, which significantly reduces the number of Gaussians and achieves real-time dynamic scene rendering by modeling time-varying attributes and automatically filtering static anchors using a temporal masking MLP.
Efficient Graph Bandit Learning with Side-Observations and Switching Constraints
Xueping Gong (Hong Kong University of Science and Technology), Jiheng Zhang (Hong Kong University of Science and Technology)
OptimizationGraph Neural NetworkReinforcement LearningGraph
🎯 What it does: A multi-armed bandit framework with graphical feedback and graphical constraints is proposed and studied, and algorithms for optimal learning are designed, both graph-independent and graph-dependent.
Efficient Image-to-Image Diffusion Classifier for Adversarial Robustness
Hefei Mei (City University of Hong Kong), Chang Xu (University of Sydney)
ClassificationComputational EfficiencyAdversarial AttackConvolutional Neural NetworkDiffusion modelImage
🎯 What it does: This study proposes an Image-to-Image Diffusion Classifier (IDC) that achieves efficient adversarial robustness by transforming the classification task into image label translation.
Efficient Indoor Depth Completion Network Using Mask-adaptive Gated Convolution
Tingxuan Huang (Northeastern University), Dongyue Chen (Northeastern University)
RestorationDepth EstimationConvolutional Neural NetworkPoint Cloud
🎯 What it does: This paper proposes an indoor depth completion network based on Mask-adaptive Gated Convolution (MagaConv) and Bidirectional Aligned Projection (Bid-AP) to fill in the missing areas of depth maps generated by sensors such as TOF and structured light.
Efficient Language-instructed Skill Acquisition via Reward-Policy Co-Evolution
Changxin Huang (Shenzhen University), Jianqiang Li (Shenzhen MSU-BIT University)
OptimizationRobotic IntelligenceReinforcement Learning from Human FeedbackTransformerLarge Language ModelReinforcement LearningSequential
🎯 What it does: This paper proposes a Reward-Policy Co-evolution Framework (ROSKA) that enables large language models (LLMs) to generate reward functions and co-evolve with policies in each round of co-evolution, thereby achieving efficient learning of high-dimensional robotic skills through language instructions.
Efficient Multi-Policy Evaluation for Reinforcement Learning
Shuze Daniel Liu (University of Virginia), Shangtong Zhang (University of Virginia)
Reinforcement LearningTabular
🎯 What it does: An unbiased and efficient method for multi-objective policy evaluation is proposed, which achieves sample sharing by designing a unified behavior policy for all target policies, significantly reducing variance.
Efficient Neural Network Encoding for 3D Color Lookup Tables
Vahid Zehtab (University of Toronto), Michael S. Brown (York University)
CompressionConvolutional Neural NetworkFlow-based ModelImage
🎯 What it does: A lightweight residual network combined with reversible flow is used to compress hundreds of 3D color lookup tables into a unified model of less than 0.25 MB, while keeping color distortion during reconstruction to no more than ΔE 2.
Efficient Online Training for Zero-Shot Time-Lapse Microscopy Denoising and Super-Resolution
Ruian He (Fudan University), Bo Yan (Fudan University)
RestorationSuper ResolutionConvolutional Neural NetworkImageVideo
🎯 What it does: This paper proposes the MDSR-Zero method, which achieves zero-shot online training for temporal microscopy video denoising and super-resolution, significantly reducing training time and enhancing temporal consistency.