AAAI 2025 Papers — Page 7
AAAI Conference on Artificial Intelligence · 3028 papers
Crossfire: An Elastic Defense Framework for Graph Neural Networks Under Bit Flip Attacks
Lorenz Kummer (University of Vienna), Nils Morten Kriege (University of Vienna)
Anomaly DetectionAdversarial AttackGraph Neural NetworkGraph
🎯 What it does: Proposes the Crossfire framework for detecting and recovering from Bit Flip Attacks (BFA) on GNNs, capable of restoring the model to its pre-attack state without retraining, using unlabeled data, and without posterior testing.
CROSSNEWS: A Cross-Genre Authorship Verification and Attribution Benchmark
Marcus Ma (Georgia Institute of Technology), Wei Xu (Georgia Institute of Technology)
RecognitionLarge Language ModelPrompt EngineeringTextBenchmark
🎯 What it does: A CROSSNEWS cross-genre author identification dataset was constructed, and various author identification models were evaluated, proposing a zero-shot LLM embedding method called SELMA.
CryoDomain: Sequence-free Protein Domain Identification from Low-resolution Cryo-EM Density Maps
Muzhi Dai (Tsinghua University), Qiangfeng Cliff Zhang (Tsinghua University)
RecognitionRetrievalProtein Structure PredictionContrastive LearningImage
🎯 What it does: Proposed the CryoDomain method, which identifies protein domains from low-resolution Cryo-EM density maps without the need for protein sequence information.
CSformer: Combining Channel Independence and Mixing for Robust Multivariate Time Series Forecasting
Haoxin Wang (Sichuan University), Site Mo (Sichuan University)
TransformerTime Series
🎯 What it does: Proposes CSformer, which integrates channel independence and channel mixing strategies, using a two-stage shared parameter multi-head self-attention to extract and fuse channel and sequence information;
CSL-L2M: Controllable Song-Level Lyric-to-Melody Generation Based on Conditional Transformer with Fine-Grained Lyric and Musical Controls
Li Chai (Westlake University), Donglin Wang (Westlake University)
GenerationTransformerTextAudio
🎯 What it does: A controllable full lyric-to-melody generation method called CSL-L2M is proposed, which can generate complete and structurally reasonable melodies based on lyrics and user-specified musical attributes.
CSR:Achieving 1 Bit Key-Value Cache via Sparse Representation
Hongxuan Zhang (Nanjing University), Guihai Chen (Nanjing University)
CompressionComputational EfficiencyTransformerLarge Language ModelText
🎯 What it does: Proposes the Cache Sparse Representation (CSR) scheme, which transforms the KV cache in LLM inference from dense vectors to sparse indices and weights, significantly reducing memory usage;
CSSinger: End-to-End Chunkwise Streaming Singing Voice Synthesis System Based on Conditional Variational Autoencoder
Jianwei Cui (University of Science and Technology of China), Lirong Dai (University of Science and Technology of China)
GenerationData SynthesisAuto EncoderGenerative Adversarial NetworkAudio
🎯 What it does: Designed and implemented a chunk-based streaming singing synthesis system CSSinger based on conditional VAE, and launched a fully streaming version CSSinger-FS;
CTD4 – a Deep Continuous Distributional Actor-Critic Agent with a Kalman Fusion of Multiple Critics
David Valencia (University of Auckland), Minas Liarokapis (University of Auckland)
Reinforcement LearningAgentic AISequential
🎯 What it does: A continuous distributed Actor-Critic algorithm named CTD4 is proposed, which modifies TD3 to output a normal distribution for the critic, using multiple critics fused through Kalman filtering to reduce overestimation and improve sample efficiency.
CtrlAvatar: Controllable Avatars Generation via Disentangled Invertible Networks
Wenfeng Song (Beijing Information Science and Technology University), Xia Hou (Beijing Information Science and Technology University)
GenerationData SynthesisPose EstimationFlow-based ModelMesh
🎯 What it does: Proposes the CtrlAvatar framework, which utilizes separable invertible networks to achieve decoupled generation of realistic and customizable 3D human avatars with distinct poses and textures;
CUGF: A Reliable and Fair Recommendation Framework
Nitin Bisht (University of Technology Sydney), Guandong Xu (Education University of Hong Kong)
Recommendation SystemGraph Neural NetworkTabular
🎯 What it does: A recommendation framework CUGF based on conformal prediction is proposed to generate prediction sets with adjustable probabilities of covering real items while ensuring fairness among different user groups.
Cultivating Archipelago of Forests: Evolving Robust Decision Trees Through Island Coevolution
Adam Zychowski (Warsaw University of Technology), Jacek Mańdziuk (AGH University of Krakow)
ClassificationOptimizationAdversarial AttackTabular
🎯 What it does: A collaborative evolutionary algorithm based on islands, ICoEvoRDF, is proposed to construct robust decision tree forests, capable of optimizing adversarial accuracy and minimizing maximum regret under a given perturbation radius.
CUQDS: Conformal Uncertainty Quantification Under Distribution Shift for Trajectory Prediction
Huiqun Huang (University of Connecticut), Fei Miao (University of Texas at Arlington)
Autonomous DrivingTime Series
🎯 What it does: This paper proposes the CUQDS framework, which combines Gaussian process regression with statistical confidence prediction to quantify and calibrate the output uncertainty of existing trajectory prediction models under distribution drift.
Curriculum Conditioned Diffusion for Multimodal Recommendation
Yimeng Yang (Shandong University), Xiangxu Meng (Shandong University)
Recommendation SystemGraph Neural NetworkTransformerDiffusion modelImageTextMultimodality
🎯 What it does: The CCDRec framework is proposed, utilizing conditional diffusion models to achieve multimodal information alignment, and dynamically adapting negative sample sampling during the training process.
CustomContrast: A Multilevel Contrastive Perspective for Subject-Driven Text-to-Image Customization
Nan Chen (University of Science and Technology of China), Zhendong Mao (University of Science and Technology of China)
Image TranslationGenerationTransformerDiffusion modelContrastive LearningImageText
🎯 What it does: A text-to-image customization framework called CustomContrast is proposed from a cross-differential perspective, which achieves better text controllability while maintaining high similarity.
CustomCrafter: Customized Video Generation with Preserving Motion and Concept Composition Abilities
Tao Wu (Zhejiang University), Xi Li (Zhejiang University)
GenerationData SynthesisDiffusion modelImageVideoText
🎯 What it does: The CustomCrafter framework is proposed, which can generate customized videos using subject learning while maintaining the motion generation and concept combination capabilities of the original video diffusion model, without the need for additional videos or re-fine-tuning.
CustomTTT: Motion and Appearance Customized Video Generation via Test-Time Training
Xiuli Bi (Chongqing University of Posts and Telecommunications), Bin Xiao (Meituan)
GenerationData SynthesisDiffusion modelVideo
🎯 What it does: This paper proposes the CustomTTT method, which achieves simultaneous customization of video appearance and motion within the same network.
CVLUE: A New Benchmark Dataset for Chinese Vision-Language Understanding Evaluation
Yuxuan Wang (Zhejiang Lab), Hongyang Chen (Zhejiang Lab)
RetrievalVision Language ModelImageTextMultimodalityBenchmark
🎯 What it does: This paper proposes the CVLUE benchmark for comprehensively evaluating the understanding ability of visual-language models (VLM) in the context of Chinese culture.
CyberPal.AI: Empowering LLMs with Expert-Driven Cybersecurity Instructions
Matan Levi (IBM Research), Anton Puzanov (IBM Research)
Safty and PrivacyAdversarial AttackTransformerLarge Language ModelSupervised Fine-TuningPrompt EngineeringText
🎯 What it does: This paper constructs a security instruction dataset, SecKnowledge, with approximately 400,000 entries through a two-stage process assisted by expert-defined schemas and LLMs. The dataset is used to fine-tune LLMs, resulting in the CyberPal.AI series of security expert models, and the SecKnowledge-Eval evaluation set is designed for comprehensive assessment of the models.
Cycle3D: High-quality and Consistent Image-to-3D Generation via Generation-Reconstruction Cycle
Zhenyu Tang (Peking University), Li Yuan (Peking University)
GenerationTransformerDiffusion modelGaussian SplattingImagePoint Cloud
🎯 What it does: Proposes Cycle3D, a unified image-to-3D generation framework that cyclically uses 2D diffusion generative models and 3D reconstruction models during multi-step diffusion processes;
D&M: Enriching E-commerce Videos with Sound Effects by Key Moment Detection and SFX Matching
Jingyu Liu (Renmin University of China), Xirong Li (Renmin University of China)
Object DetectionRecommendation SystemTransformerContrastive LearningVideoMultimodalityAudio
🎯 What it does: An end-to-end D&M method is proposed to automatically detect key moments in e-commerce videos and match them with appropriate sound effects (VDSFX).
D^2-DPM: Dual Denoising for Quantized Diffusion Probabilistic Models
Qian Zeng (Zhejiang University), Mingli Song (Alibaba Group)
RestorationGenerationDiffusion modelImageStochastic Differential EquationOrdinary Differential Equation
🎯 What it does: Perform dual denoising on quantized diffusion models to eliminate mean and variance bias caused by post-training quantization noise.
DABL: Detecting Semantic Anomalies in Business Processes Using Large Language Models
Wei Guan (Shanghai Jiao Tong University), Shiyou Qian (University of Shanghai for Science and Technology)
Anomaly DetectionTransformerLarge Language ModelSupervised Fine-TuningPrompt EngineeringTextSequential
🎯 What it does: DABL is a method that uses large language models to detect semantic anomalies in business processes and explains the reasons for these anomalies in natural language.
DAMMFND: Domain-Aware Multimodal Multi-view Fake News Detection
Weihai Lu (Peking University), Zhiqiu Ye (Anhui University)
ClassificationDomain AdaptationTransformerLarge Language ModelImageTextMultimodality
🎯 What it does: This paper proposes the Domain-Aware Multi-Modal Multi-View Fake News Detection (DAMMFND) framework to address issues such as inaccurate domain recognition, negative transfer, and uneven contributions of domain heterogeneous modalities in multi-modal multi-domain fake news detection.
DAMPER: A Dual-Stage Medical Report Generation Framework with Coarse-Grained MeSH Alignment and Fine-Grained Hypergraph Matching
Xiaofei Huang (Shenzhen University), Linlin Shen (Shenzhen University)
GenerationGraph Neural NetworkTransformerGenerative Adversarial NetworkImageTextMultimodalityElectronic Health Records
🎯 What it does: A dual-stage medical report generation framework called DAMPER is proposed, which first uses MeSH-guided coarse-grained alignment to extract an overall impression, and then employs hypergraph matching for fine-grained alignment to generate more accurate chest X-ray reports.
DanceFix: An Exploration in Group Dance Neatness Assessment Through Fixing Abnormal Challenges of Human Pose
Huangbiao Xu (Fuzhou University), Wenzhong Guo (Fuzhou University)
Object TrackingPose EstimationTransformerOptical FlowVideo
🎯 What it does: To address the issues of occlusion and motion blur that lead to abnormal keypoints in group dance pose estimation, a system called DanceFix is proposed. It corrects abnormal skeleton points through Bidirectional Optical Flow Correction (BOFC) and Task-Split-based Part Motion Feature Extraction (PETD). Based on this, a Group Dance Neatness Assessment (GDNA) algorithm is designed to quantitatively evaluate the neatness of dance movements.
DAPoinTr: Domain Adaptive Point Transformer for Point Cloud Completion
Yinghui Li (Deakin University), Xuequan Lu (University of Western Australia)
Domain AdaptationTransformerPoint Cloud
🎯 What it does: Proposes a Domain Adaptive Point Transformer (DAPoinTr) framework for cross-domain point cloud completion.
DARR: A Dual-Branch Arithmetic Regression Reasoning Framework for Solving Machine Number Reasoning
Chengtai Li (University of Nottingham Ningbo China), Xudong Jiang (Nanyang Technological University)
ClassificationRecognitionConvolutional Neural NetworkImage
🎯 What it does: A dual-branch arithmetic regression reasoning framework (DARR) is proposed to address the machine number reasoning (MNR) task, which involves identifying candidate answer images that follow the same arithmetic rules among three contextual images.
DASK: Distribution Rehearsing via Adaptive Style Kernel Learning for Exemplar-Free Lifelong Person Re-Identification
Kunlun Xu (Peking University), Jiahuan Zhou (Peking University)
RecognitionRetrievalConvolutional Neural NetworkContrastive LearningImage
🎯 What it does: The DASK method is proposed, utilizing distribution replay and adaptive style kernel learning to achieve zero-shot lifelong person re-identification, avoiding the storage of historical data.
Data Augmentation for Instruction Following Policies via Trajectory Segmentation
Niklas Hoepner (University of Amsterdam), Herke van Hoof (Vrije Universiteit Amsterdam)
SegmentationRobotic IntelligenceReinforcement Learning from Human FeedbackReinforcement LearningVideo
🎯 What it does: Using a limited number of labeled sub-trajectories to segment unlabeled long trajectories and assign instructions to each segment, thereby providing more training samples for instruction-following strategies.
Data-Free Black-Box Federated Learning via Zeroth-Order Gradient Estimation
Xinge Ma (Yunnan University), Xuejie Zhang (Yunnan University)
Data SynthesisFederated LearningKnowledge DistillationAdversarial AttackGenerative Adversarial NetworkImage
🎯 What it does: This paper proposes FedZGE, a data-independent, black-box federated learning framework that trains a generator on the server side to generate synthetic data for knowledge distillation, without the need to share model parameters or auxiliary data throughout the process.
Data-Free Universal Attack by Exploiting the Intrinsic Vulnerability of Deep Models
YangTian Yan (Macau University of Science and Technology), Jinyu Tian (Macau University of Science and Technology)
Adversarial AttackConvolutional Neural NetworkImage
🎯 What it does: A data-free universal adversarial perturbation generation method called IntriUAP is proposed, which utilizes the inherent vulnerability of linear layers in deep networks to generate attack perturbations.
DC-PCN: Point Cloud Completion Network with Dual-Codebook Guided Quantization
Qiuxia Wu (South China University of Technology), Kun Hu (The University of Sydney)
GenerationData SynthesisDomain AdaptationTransformerContrastive LearningPoint Cloud
🎯 What it does: A Dual-Codebook Point Completion Network (DC‑PCN) is proposed, achieving complete reconstruction of point clouds through an encoder-decoder architecture.
DCA: Dividing and Conquering Amnesia in Incremental Object Detection
Aoting Zhang (Institute of Information Engineering Chinese Academy of Sciences), Yu Zhou (Nankai University)
Object DetectionKnowledge DistillationTransformerLarge Language ModelImage
🎯 What it does: This paper proposes an incremental object detection method based on Transformer, called DCA, which separates localization from recognition and utilizes the semantic features of a pre-trained language model to guide recognition, while employing a dual classifier fusion to alleviate recognition forgetting.
DCC: Differentiable Cardinality Constraints for Partial Index Tracking
Wooyeon Jo (Ajou University), Hyunsouk Cho (Ajou University)
OptimizationComputational EfficiencyTime SeriesFinance Related
🎯 What it does: This study investigates how to solve the partial replication problem in index tracking using differentiable cardinality constraints.
DCHM: Dynamic Collaboration of Heterogeneous Models Through Isomerism Learning in a Blockchain-Powered Federated Learning Framework
Zhihao Hao (Beijing Technology and Business University), Haisheng Li (Beijing Technology and Business University)
OptimizationFederated LearningTime Series
🎯 What it does: This paper proposes a distributed heterogeneous model collaborative framework DCHM based on isomerism learning, utilizing alliance chains to achieve trustworthy and decentralized model collaboration and result aggregation.
DCILP: A Distributed Approach for Large-Scale Causal Structure Learning
Shuyu Dong (INRIA), Koji Maruhashi (Fujitsu Limited)
Graph Neural NetworkGraphTabularBiomedical Data
🎯 What it does: A divide-and-conquer causal structure learning framework DCILP is proposed, which first estimates the Markov blanket of each variable and solves local subproblems in parallel, and then unifies these local results using integer linear programming (ILP) to obtain a global causal graph.
DCSF-KD: Dynamic Channel-wise Spatial Feature Knowledge Distillation for Object Detection
Tao Dai (Shenzhen University), Zexuan Zhu (Shenzhen University)
Object DetectionKnowledge DistillationImage
🎯 What it does: A dynamic channel-space feature distillation framework DCSF-KD is proposed for knowledge distillation in object detection, which utilizes channel weights to dynamically extract spatial features for distillation.
DCTMamba: Advancing JPEG Image Restoration Through Long-Sequence Modeling and Adaptive Frequency Strategy
Xi Wang (University of Science and Technology of China), Zheng-Jun Zha (University of Science and Technology of China)
RestorationCompressionConvolutional Neural NetworkImage
🎯 What it does: In the JPEG image restoration task, the DCTMamba framework is proposed, which combines discrete cosine transform with Mamba to achieve causal scanning from low frequency to high frequency.
DDJND: Dual Domain Just Noticeable Difference in Multi-Source Content Images with Structural Discrepancy
Miaohui Wang (Shenzhen University), Wuyuan Xie (Shenzhen University)
Knowledge DistillationConvolutional Neural NetworkImage
🎯 What it does: A dual-domain (spatial and frequency domain) joint modeling framework for JND (Just Noticeable Difference) estimation is proposed, achieving more accurate visual redundancy prediction for structural differences in multi-source content images.
De-singularity Subgradient for the q-th-Powered lₚ-Norm Weber Location Problem
Zhao-Rong Lai (Jinan University), Cheng Li (Jinan University)
OptimizationTime SeriesFinance Related
🎯 What it does: A de-singular subgradient method is proposed for the q-th powered ℓp-norm Weber location problem for 1 ≤ p < 2 and 1 ≤ q ≤ p, along with the corresponding q-P-NWAWS algorithm.
DearLLM: Enhancing Personalized Healthcare via Large Language Models-Deduced Feature Correlations
Yongxin Xu (Peking University), Bing Xie (Peking University)
ClassificationRecommendation SystemGraph Neural NetworkTransformerLarge Language ModelTabularBiomedical DataElectronic Health Records
🎯 What it does: This paper proposes a framework named DearLLM, which utilizes large language models (LLM) to derive quantitative associations between diagnostic codes in the context of personalized medicine, constructs a feature prior graph, and enhances EHR prediction performance through graph convolutional networks (GCN) and frequency-aware graph pooling.
Debate on Graph: A Flexible and Reliable Reasoning Framework for Large Language Models
Jie Ma (Xi'an Jiaotong University), Lizhen Cui (Northwestern Polytechnical University)
TransformerLarge Language ModelPrompt EngineeringTextGraph
🎯 What it does: An iterative interactive framework named DoG is proposed, which utilizes large language models for subgraph focusing and multi-role debate on knowledge graphs, gradually reasoning and generating answers.
Debiased Active Learning with Variational Gradient Rectifier
Weiguo Chen (National University of Defense Technology), Shanshan Li (Tianjin University)
ClassificationData-Centric LearningAuto EncoderImageText
🎯 What it does: This paper proposes an active learning framework based on Variational Gradient Corrector (VaGeRy) to measure and dynamically correct biases during the training process, enhancing the sample selection effectiveness of active learning.
Debiased All-in-one Image Restoration with Task Uncertainty Regularization
Gang Wu (Harbin Institute of Technology), Xianming Liu (Harbin Institute of Technology)
RestorationConvolutional Neural NetworkImage
🎯 What it does: An adaptive multi-task regularization method based on task uncertainty (Task Uncertainty Regularization, TUR) is proposed, which automatically learns task weights during joint training of all tasks and improves recovery quality.
Debiased Distillation for Consistency Regularization
Lu Wang (Northeastern University), Jun Cheng (Shenzhen Institute of Advanced Technology)
ClassificationKnowledge DistillationImage
🎯 What it does: This paper proposes a new knowledge distillation strategy—Intra-Class Knowledge Distillation (IKD), which achieves prediction consistency by sharing the average logits within the same class, reducing noise and bias, and enhancing the generalization ability of the student model.
Debiased Multimodal Understanding for Human Language Sequences
Zhi Xu (Fudan University), Lihua Zhang (Fudan University)
ClassificationRecognitionDomain AdaptationTransformerSupervised Fine-TuningContrastive LearningTextMultimodalityAudio
🎯 What it does: A subjective factor debiasing module based on causal intervention, SuCI, is proposed to eliminate prediction bias caused by subject differences in multimodal language understanding (MLU), and this module can be directly embedded as a general plugin into existing MLU models.
Decentralized and Uncoordinated Learning of Stable Matchings: A Game-Theoretic Approach
S. Rasoul Etesami (University of Illinois Urbana-Champaign), R. Srikant (University of Illinois Urbana-Champaign)
OptimizationReinforcement Learning
🎯 What it does: Learning stable matching in a decentralized uncoordinated environment
Decentralized Convergence to Equilibrium Prices in Trading Networks
Edwin Lock (University of Oxford), Paul W. Goldberg (University of Oxford)
OptimizationGraph Neural NetworkGraphFinance Related
🎯 What it does: A decentralized market dynamic based on best responses is proposed, where agents update their bids through bilateral negotiations, ultimately converging to a competitive equilibrium.
Decentralized Federated Learning with Model Caching on Mobile Agents
Xiaoyu Wang (New York University), Yong Liu (New York University)
Federated LearningImage
🎯 What it does: A decentralized federated learning framework called Cached-DFL is proposed, which utilizes D2D communication and caching mechanisms between mobile agents to achieve delay-tolerant model propagation and aggregation.
Decentralized Projected Riemannian Stochastic Recursive Momentum Method for Nonconvex Optimization
Kangkang Deng (National University of Defense Technology), Jiang Hu (University of California)
Optimization
🎯 What it does: This paper studies a decentralized optimization method aimed at jointly minimizing the sum of local non-convex cost functions through real-time processing at n nodes in the network.
DECIDER: Difference-aware Contrastive Diffusion Model with Adversarial Perturbations for Image Change Captioning
Guojin Zhong (Hunan University), Wenbo Pan (CRRC Zhuzhou Institute)
RecognitionGenerationTransformerDiffusion modelContrastive LearningImageText
🎯 What it does: By constructing a difference-aware contrastive diffusion model and incorporating adversarial perturbations, natural language descriptions of subtle changes between two similar images are achieved.
Decomposed Quadratization: Efficient QUBO Formulation for Learning Bayesian Network
Yuta Shikuri (Tokio Marine Holdings)
OptimizationTabular
🎯 What it does: This paper proposes a quadratic unconstrained binary optimization (QUBO) model based on decomposed quadratization and candidate parent subsets for efficiently learning Bayesian network structures.
Decomposed Spatio-Temporal Mamba for Long-Term Traffic Prediction
Sicheng He (Beijing University of Technology), Minglong Lei (Beijing University of Technology)
Graph Neural NetworkTime Series
🎯 What it does: A Decomposed Spatio-Temporal Mamba model (DST-Mamba) is proposed for long-term traffic flow prediction.
Decomposing and Fusing Intra- and Inter-Sensor Spatio-Temporal Signal for Multi-Sensor Wearable Human Activity Recognition
Haoyu Xie (Chongqing University), Li Liu (Chongqing University)
RecognitionConvolutional Neural NetworkSupervised Fine-TuningMultimodalityTime Series
🎯 What it does: A decomposition and fusion framework called DecomposeWHAR is proposed to better capture the internal and external spatiotemporal relationships in multi-sensor wearable human action recognition.
Deconfound Semantic Shift and Incompleteness in Incremental Few-shot Semantic Segmentation
Yirui Wu (Hohai University), Shaohua Wan (University of Electronic Science and Technology of China)
SegmentationKnowledge DistillationMeta LearningImage
🎯 What it does: This study addresses the issues of semantic drift and incompleteness in incremental few-shot semantic segmentation, proposing a causal framework and designing the CIM and PRM modules for debiasing and semantic completion.
DeCorrNet: Enhancing Neural Decoding Performance by Eliminating Correlations in Noise
Xianhan Tan (Zhejiang University), Yueming Wang (Zhejiang University)
ClassificationRecognitionOptimizationRecurrent Neural NetworkTransformerTime SeriesBiomedical Data
🎯 What it does: A denoising network called DeCorrNet based on an attention mechanism is proposed, specifically designed to eliminate noise correlation in neural signals, thereby enhancing neural decoding performance.
Decoupled Policy Actor-Critic: Bridging Pessimism and Risk Awareness in Reinforcement Learning
Michal Nauman (University of Warsaw), Marek Cygan (Nomagic)
Reinforcement LearningSequential
🎯 What it does: A Decoupled Policy Actor-Critic (DAC) algorithm is proposed, utilizing two separate actor networks to handle optimistic exploration and pessimistic temporal difference learning, thereby improving sample efficiency and final performance.
Decoupled Spatio-Temporal Consistency Learning for Self-Supervised Tracking
Yaozong Zheng (Guangxi Normal University), Shuxiang Song (Guangxi Normal University)
Object TrackingTransformerContrastive LearningVideo
🎯 What it does: A frame-free annotation self-supervised visual tracking framework called SSTrack is developed, which learns cross-frame object representations through separated spatiotemporal consistency training and instance contrastive loss.
Decoupling Appearance Variations with 3D Consistent Features in Gaussian Splatting
Jiaqi Lin (Tsinghua University), Wenming Yang (Huawei Noah's Ark Lab)
GenerationOptimizationGaussian SplattingImage
🎯 What it does: A method named DAVIGS is proposed for appearance decoupling, aimed at eliminating floating points and color distortions caused by camera ISP, time, weather, and local lighting changes within the Gaussian Splatting framework, thereby achieving high-quality, real-time novel view rendering.
Decoupling Metacognition from Cognition: A Framework for Quantifying Metacognitive Ability in LLMs
Guoqing Wang (East China Normal University), Hong Zheng (East China Normal University)
TransformerLarge Language ModelText
🎯 What it does: Proposed and validated the DMC framework for quantifying the metacognitive abilities of large language models (LLMs) and decoupling them from cognitive abilities.
Decoupling Scattering: Pseudo-Label Guided NeRF for Scenes with Scattering Media
Mingyang Zhang (East China Normal University), Guixu Zhang (East China Normal University)
RestorationGenerationOptimizationNeural Radiance FieldImage
🎯 What it does: In scattering medium (fog/underwater) scenarios, a pseudo-label guided NeRF model is proposed, combined with a cyclic progressive dimension optimization strategy (CPDOS) to achieve the separation and reconstruction of object and medium density.
DECT: Harnessing LLM-assisted Fine-Grained Linguistic Knowledge and Label-Switched and Label-Preserved Data Generation for Diagnosis of Alzheimer’s Disease
Tingyu Mo (University of Hong Kong), Lawrence Y. L. Cheung (Chinese University of Hong Kong)
ClassificationData SynthesisTransformerLarge Language ModelPrompt EngineeringTextBiomedical DataAlzheimer's DiseaseAudio
🎯 What it does: DECT was developed, a fine-grained language knowledge extraction and label switching/retention data generation framework based on large language models, aimed at improving the accuracy of speech diagnosis for Alzheimer's disease.
Deep Disentangled Metric Learning
Jinhee Park (Chung Ang University), Junseok Kwon (Chung Ang University)
ClassificationRetrievalContrastive LearningImage
🎯 What it does: Introducing class-agnostic regularization in deep metric learning, utilizing the information bottleneck framework to learn decoupled representations of class-specific and class-agnostic features, and constructing the DDML module;
Deep Evidential Hashing for Trustworthy Cross-Modal Retrieval
Yuan Li (Sichuan University), Peng Hu
RetrievalMultimodality
🎯 What it does: This paper proposes a Deep Evidential Cross-Modal Hashing (DECH) method that utilizes evidence learning to directly generate binary codes and provides reliability assessment for retrieval results.
Deep Generative Model for Mechanical System Configuration Design
Yasaman Etesam (Simon Fraser University), Pradeep Kumar Jayaraman (Autodesk Research)
GenerationOptimizationTransformerSequential
🎯 What it does: This paper proposes a Transformer-based deep generative model called GearFormer, which is used to generate configurations of gear transmission systems that meet multiple design constraints in a single pass, and further combines it with traditional search methods (EDA, MCTS) to form a hybrid design scheme.
Deep Graph Online Hashing for Multi-Label Image Retrieval
Yuan Cao (Ocean University of China), Jie Gui (Southeast University)
RetrievalGraph Neural NetworkImage
🎯 What it does: This paper proposes and implements Deep Graph Online Hashing (DGOH), achieving online deep hashing in multi-label image retrieval scenarios while balancing retrieval accuracy and training efficiency.
Deep Hierarchies and Invariant Disease-Indicative Feature Learning for Computer Aided Diagnosis of Multiple Fundus Diseases
Yuxin Lin (Harbin Institute of Technology), Yong Xu (Harbin Institute of Technology)
ClassificationConvolutional Neural NetworkTransformerContrastive LearningImage
🎯 What it does: This paper proposes a hierarchical visual transformer (HVT) that integrates multi-level features and a self-supervised mask-consistent feature learner (MCFL) for the automatic diagnosis of multiple retinal diseases.
Deep Hypergraph Neural Networks with Tight Framelets
Ming Li (Zhejiang Normal University), Pietro Liò (Cambridge University)
Graph Neural NetworkGraph
🎯 What it does: This study investigates the oversmoothing problem in deep hypergraph neural networks and proposes the FrameHGNN model based on tight framelets.
Deep Implicit Imitation Reinforcement Learning in Heterogeneous Action Settings
Iason Chrysomallis (Technical University of Crete), Markos Papageorgiou (Technical University of Crete)
Reinforcement Learning
🎯 What it does: Proposes HA-DIIQN, an extension of DIIQN to achieve implicit imitation reinforcement learning in heterogeneous environments where the action sets of the mentor and observer are inconsistent;
Deep Multi-modal Graph Clustering via Graph Transformer Network
Qianqian Wang (Xidian University), Quanxue Gao (Xidian University)
Graph Neural NetworkTransformerMultimodalityGraph
🎯 What it does: A deep multimodal graph clustering algorithm based on Graph Transformer Networks (DMGC-GTN) is proposed, which utilizes various embeddings such as graph Laplacian smoothing, node proximity, and shortest paths to obtain a unified low-dimensional representation for direct clustering.
Deep Non-Rigid Structure-from-Motion Revisited: Canonicalization and Sequence Modeling
Hui Deng (Northwestern Polytechnical University), Yuchao Dai (Northwestern Polytechnical University)
Pose EstimationConvolutional Neural NetworkVideoTime SeriesSequential
🎯 What it does: A sequence-based Canonicalization (GPA layer) and sequence modeling deep NRSfM framework is proposed, achieving the reconstruction of 3D non-rigid shape sequences from 2D sequences.
Deep Rank-One Tensor Functional Factorization for Multi-Dimensional Data Recovery
Yanyi Li (Xi'an Jiaotong University), Deyu Meng (Xi'an Jiaotong University)
RestorationCompressionAuto EncoderImageVideo
🎯 What it does: This paper proposes a Deep Rank-One Tensor Function Decomposition (DRO-TFF) framework for efficiently recovering multi-dimensional data such as color images, videos, and hyperspectral images.
Deep Reinforcement Learning with Time-Scale Invariant Memory
Md Rysul Kabir (Indiana University), Zoran Tiganj (Indiana University)
Recurrent Neural NetworkReinforcement LearningTime Series
🎯 What it does: This paper embeds a scale-invariant memory model based on computational neuroscience into deep reinforcement learning agents and validates its effectiveness on various interval timing and memory tasks.
Deep Representation Learning for Forecasting Recursive and Multi-Relational Events in Temporal Networks
Tony Gracious (Indian Institute of Science), Ambedkar Dukkipati (Indian Institute of Science)
Representation LearningGraph Neural NetworkTransformerContrastive LearningGraphTime SeriesOrdinary Differential Equation
🎯 What it does: This paper proposes a time point process model based on recursive multi-relation hyperedges, called RRHyperTPP, for predicting the types and occurrence times of complex high-order interaction events.
Deep Submodular Optimization and LLM for Multimodal Content Extraction and Automatic Poster Generation from Long Document
Vijay Jaisankar (International Institute of Information Technology), Shwetha Somasundaram (Adobe Research)
GenerationOptimizationTransformerLarge Language ModelImageTextMultimodality
🎯 What it does: This paper presents an end-to-end system called PostDoc, which automatically converts long multimodal documents (text + images) into single-page posters, covering content extraction, LLM paraphrasing, and template generation.
Deep-Union Completion
Siddharth Baskar (Wisconsin Institute for Discovery), Daniel L. Pimentel-Alarcón
RestorationAuto EncoderTabular
🎯 What it does: This paper proposes a Deep Union Completion (DUC) framework that utilizes a pseudo-completion layer to reconstruct data with a high proportion of missing values, and simultaneously performs missing value imputation and subspace clustering through an autoencoder and a closed clustering layer.
DeepSN: A Sheaf Neural Framework for Influence Maximization
Asela Hevapathige (Australian National University), Ahad N. Zehmakan (Australian National University)
OptimizationGraph Neural NetworkGraph
🎯 What it does: A deep neural framework called DeepSN based on sheaf theory is proposed for learning influence diffusion models and optimizing seed selection to achieve maximum influence.
DEEPTalk: Dynamic Emotion Embedding for Probabilistic Speech-Driven 3D Face Animation
Jisoo Kim (Yonsei University), Youngjae Yu (Yonsei University)
GenerationContrastive LearningMultimodalityAudio
🎯 What it does: This paper proposes a voice-based 3D talking head animation system called DEEPTalk, which can generate diverse and emotionally rich facial expressions based on input speech.
Defeasible Visual Entailment: Benchmark, Evaluator, and Reward-Driven Optimization
Yue Zhang (University of Texas at Dallas), Vibhav Gogate (University of Texas at Dallas)
GenerationOptimizationTransformerLarge Language ModelContrastive LearningImageTextMultimodalityBenchmark
🎯 What it does: This paper proposes the Defeasible Visual Entailment (DVE) task, the first corresponding benchmark dataset, a no-reference reasoning perception evaluator based on contrastive learning, and a reward-driven update optimization framework.
Defending Against Sophisticated Poisoning Attacks with RL-based Aggregation in Federated Learning
Yujing Wang (Beihang University), Binghui Guo (Beihang University)
Federated LearningReinforcement LearningImage
🎯 What it does: Proposes an adaptive aggregation method AdaAggRL based on reinforcement learning to defend against advanced poisoning attacks targeting servers.
Defense Against Model Stealing Based on Account-Aware Distribution Discrepancy
Jian-Ping Mei (Zhejiang University of Technology), Tiantian Zhu (Macquarie University)
ClassificationAnomaly DetectionImage
🎯 What it does: A non-parametric detector based on Account-Aware Distribution Difference (ADD) is proposed, which is combined with random label poisoning to form an end-to-end real-time defense module D-ADD, aimed at preventing model stealing attacks on image classification models.
Dehaze-RetinexGAN: Real-World Image Dehazing via Retinex-based Generative Adversarial Network
Xinran Wang (Southwest University), Yun Liu (Southwest University)
RestorationGenerationGenerative Adversarial NetworkImage
🎯 What it does: A lightweight Dehaze-RetinexGAN is proposed, achieving real scene image dehazing under unpaired data through a two-stage process of self-supervised pre-training and weakly supervised fine-tuning.
Delay as Payoff in MAB
Ofir Schlisselberg (Tel Aviv University), Yishay Mansour (Tel Aviv University)
Recommendation SystemOptimizationReinforcement LearningTabular
🎯 What it does: This paper studies a variant of the classic stochastic multi-armed bandit (MAB) problem, where the rewards (costs or rewards) obtained by the agent are delayed and directly related to the size of the delay. This setting realistically simulates many real-world scenarios, such as the time required for packets to be transmitted over a network or the time users spend on a webpage.
DELTA: Pre-Train a Discriminative Encoder for Legal Case Retrieval via Structural Word Alignment
Haitao Li (Tsinghua University), Yiqun Liu (Tsinghua University)
RetrievalTransformerContrastive LearningText
🎯 What it does: A pre-training framework called DELTA based on structural word alignment is proposed to enhance the discriminative ability of legal case retrieval.
DeMo: Decoupled Feature-Based Mixture of Experts for Multi-Modal Object Re-Identification
Yuhao Wang (Dalian University of Technology), Pingping Zhang (Dalian University of Technology)
RecognitionRetrievalTransformerMixture of ExpertsMultimodality
🎯 What it does: The DeMo framework is proposed, which first uses the Patch-Integrated Feature Extractor (PIFE) to extract multi-scale features, then decouples the multi-modal features hierarchically through the Hierarchical Decoupling Module (HDM), and finally applies the Attention-Triggered Mixture of Experts (ATMoE) to dynamically weight the decoupled features for multi-modal object ReID.
DeMo: Deep Motion Field Consensus with Learnable Kernels for Two-view Correspondence Learning
Yifan Lu (Wuhan University), Jiayi Ma (Chinese University of Hong Kong)
Pose EstimationOptimizationSimultaneous Localization and MappingOptical FlowImage
🎯 What it does: The DeMo network is proposed for outlier removal in two-view correspondence learning, utilizing global motion consensus to enhance matching quality.
DeNC: Unleash Neural Codecs in Video Streaming with Diffusion Enhancement
Qihua Zhou (Shenzhen University), Song Guo (Hong Kong University of Science and Technology)
GenerationCompressionDiffusion modelVideo
🎯 What it does: This paper proposes the Diffusion-enhanced Neural Codec (DeNC), which simultaneously reduces frame resolution and color bit depth through the encoder, while the decoder employs a diffusion model for perceptually guided restoration, achieving a balance among rate, distortion, and perception at extremely low bit rates for video streaming.
Denoising Diffusion Variational Inference: Diffusion Models as Expressive Variational Posteriors
Wasu Top Piriyakulkij (Cornell University), Volodymyr Kuleshov (Cornell University)
GenerationData SynthesisDiffusion modelAuto EncoderImageBiomedical Data
🎯 What it does: A new variational inference framework called DDVI is proposed, which uses diffusion models as flexible approximate posteriors to approximate the posterior distribution of latent variable models.
Dense Audio-Visual Event Localization Under Cross-Modal Consistency and Multi-Temporal Granularity Collaboration
Ziheng Zhou (Hefei University of Technology), Dan Guo (Hefei University of Technology)
TransformerVideoMultimodalityAudio
🎯 What it does: This paper addresses the task of dense audio-visual event localization in long untrimmed videos by proposing the CCNet network, which integrates cross-modal consistency collaboration and multi-temporal granularity collaboration.
Densely Connected Parameter-Efficient Tuning for Referring Image Segmentation
Jiaqi Huang (Tsinghua University), Xiu Li (Tsinghua University)
Object DetectionSegmentationTransformerContrastive LearningImageMultimodality
🎯 What it does: The DETRIS framework is proposed, utilizing a Dense Aligner and text adapter to achieve parameter-efficient fine-tuning for the Referring Image Segmentation task.
Densest k-Subgraph Mining via a Provably Tight Relaxation
Qiheng Lu (University of Virginia), Aritra Konar (KU Leuven)
OptimizationGraph Neural NetworkGraph
🎯 What it does: The paper addresses the Densest k-Subgraph problem and proposes a continuous relaxation obtained by diagonal loading (λ=1) of the adjacency matrix, proving that it converges to an integer solution when the size of the subgraph does not exceed the size of the maximum clique.
DepMGNN: Matrixial Graph Neural Network for Video-based Automatic Depression Assessment
Zijian Wu (Nanjing University of Science and Technology), Siyang Song (Nanjing University of Science and Technology)
ClassificationRecognitionGraph Neural NetworkVideo
🎯 What it does: This paper proposes a Matrix Graph Neural Network (MGNN) that directly constructs a matrix graph using the 2D feature maps of each frame in variable-length videos, enabling end-to-end learning of depression-related spatiotemporal features in full-length videos.
Depth-Centric Dehazing and Depth-Estimation from Real-World Hazy Driving Video
Junkai Fan (Nanjing University of Science and Technology), Jian Yang (Nanjing University of Science and Technology)
RestorationDepth EstimationAutonomous DrivingGenerative Adversarial NetworkVideo
🎯 What it does: This paper proposes a deep center learning framework that jointly utilizes atmospheric scattering models and brightness consistency constraints to achieve dehazing and depth estimation of monocular hazy videos.
DepthFM: Fast Generative Monocular Depth Estimation with Flow Matching
Ming Gui (CompVis at Ludwig Maximilian University of Munich), Björn Ommer (CompVis at Ludwig Maximilian University of Munich)
GenerationDepth EstimationDiffusion modelFlow-based ModelAuto EncoderImage
🎯 What it does: This paper proposes DepthFM, which utilizes flow matching technology to achieve direct generation from images to depth maps, significantly reducing sampling steps.
DEQA: Descriptions Enhanced Question-Answering Framework for Multimodal Aspect-Based Sentiment Analysis
Zhixin Han (Nankai University), Bitong Luo (JD AI Research)
ClassificationTransformerLarge Language ModelPrompt EngineeringMixture of ExpertsImageTextMultimodality
🎯 What it does: This study proposes a multimodal emotion analysis framework named DEQA, which utilizes GPT-4 to generate image descriptions and inputs text, images, and descriptions into different experts separately, completing multimodal aspect-based sentiment analysis (MABSA) through a multi-round question-and-answer approach.
DeRainGS: Gaussian Splatting for Enhanced Scene Reconstruction in Rainy Environments
Shuhong Liu (University of Tokyo), Mingrui Li (Dalian University of Technology)
RestorationTransformerGaussian SplattingImage
🎯 What it does: The paper proposes a new task for 3D scene reconstruction in rainy environments, called 3DRRE, and implements the DeRainGS method based on 3D Gaussian splatting.
Descriptive and Discriminative Document Identifiers for Generative Retrieval
Jiehan Cheng (Renmin University of China), Xiaoxi Li (Renmin University of China)
GenerationRetrievalTransformerContrastive LearningText
🎯 What it does: A learnable document identifier (D2-DocID) has been designed that can describe document semantics and distinguish similar documents, along with a supporting generative retrieval model D2Gen, which can iteratively generate better DocIDs after understanding the documents.
Design Principle Transfer in Neural Architecture Search via Large Language Models
Xun Zhou (Hong Kong Polytechnic University), Kay Chen Tan (Hong Kong Polytechnic University)
Neural Architecture SearchLarge Language ModelPrompt EngineeringImage
🎯 What it does: This paper proposes the use of large language models (LLM) to automatically learn the design principles of existing networks and apply these principles to prune the search space, thereby accelerating and improving the effectiveness of neural architecture search (NAS) on new tasks.
DesignEdit: Unify Spatial-Aware Image Editing via Training-free Inpainting with a Multi-Layered Latent Diffusion Framework
Yueru Jia (Peking University), Shanghang Zhang (Microsoft)
Image TranslationRestorationDiffusion modelImage
🎯 What it does: Achieve untrained background inpainting through key masking self-attention, and implement multi-object spatial editing within a multi-layer latent decomposition and fusion framework;
Designing Ambiguity Sets for Distributionally Robust Optimization Using Structural Causal Optimal Transport
Ahmad Reza Ehyaei, Samira Samadi (Max Planck Institute for Intelligent Systems)
OptimizationTabular
🎯 What it does: This paper proposes Structural Causal Optimal Transport (SCOT) and its relaxed version to construct a distribution uncertainty set that better aligns with real causal structures, and provides efficient solving algorithms.