AAAI 2025 Papers — Page 4
AAAI Conference on Artificial Intelligence · 3028 papers
Benchmarking and Understanding Compositional Relational Reasoning of LLMs
Ruikang Ni (Beijing University of Posts and Telecommunications), Hongliang Liang (Beijing University of Posts and Telecommunications)
TransformerLarge Language ModelTextBenchmark
🎯 What it does: A new synthetic benchmark called Generalized Associative Recall (GAR) was designed and evaluated, systematically analyzing the performance of LLMs in Combinatorial Relation Reasoning (CRR), and discovering core circuits and true/false heads through mechanism explanations.
BERT-Based Code Learning for Exception Localization and Type Prediction
Chongyu Zhang (East China Normal University), Min Zhang (East China Normal University)
Anomaly DetectionAI Code AssistantRecurrent Neural NetworkTransformerLarge Language ModelSupervised Fine-TuningText
🎯 What it does: This paper proposes the CodeHunter model, which utilizes BERT for semantic representation of code and combines Bi-LSTM for sequence labeling of code lines, achieving anomaly code localization and anomaly type prediction, along with the implementation of an IDE plugin.
Better Understandings and Configurations in MaxSAT Stochastic Local Search Solvers via Anytime Performance Analysis
Furong Ye (Chinese Academy of Sciences), Shaowei Cai (Chinese Academy of Sciences)
OptimizationHyperparameter SearchTabular
🎯 What it does: This study is the first to use the empirical cumulative distribution function (ECDF) to evaluate the 'anytime' performance of MaxSAT's stochastic local search (SLS) solvers and to improve hyperparameter configurations based on the evaluation results.
BEV-TSR: Text-Scene Retrieval in BEV Space for Autonomous Driving
Tao Tang (Shenzhen Campus of Sun Yat-sen University), Yang Wang (Li Auto Inc.)
RetrievalAutonomous DrivingTransformerLarge Language ModelContrastive LearningTextMultimodality
🎯 What it does: Proposes the BEV-TSR framework to achieve text-scene retrieval in bird's-eye view space and constructs a multi-level nuScenes-Retrieval dataset.
BEVSync: Asynchronous Data Alignment for Camera-based Vehicle-Infrastructure Cooperative Perception Under Uncertain Delays
Wentao Wang (Sun Yat-sen University), Guang Tan (Sun Yat-sen University)
Object DetectionAutonomous DrivingConvolutional Neural NetworkAuto EncoderVideo
🎯 What it does: A camera-based vehicle-road collaborative perception framework BEVSync has been designed and implemented to address the problem of asynchronous data alignment between vehicles and infrastructure in the presence of uncertain delays, by predicting synchronized features through extracting motion information from historical frames and performing BEV fusion.
Beyond Accuracy: On the Effects of Fine-Tuning Towards Vision-Language Model’s Prediction Rationality
Qitong Wang (University of Delaware), Xi Peng (University of Delaware)
ClassificationExplainability and InterpretabilityTransformerSupervised Fine-TuningVision Language ModelImageMultimodality
🎯 What it does: This study investigates the impact of mainstream VLM fine-tuning methods on prediction rationality (based on the credibility of evidence) and proposes two new metrics: prediction credibility and inference reliability.
Beyond Federated Prototype Learning: Learnable Semantic Anchors with Hyperspherical Contrast for Domain-Skewed Data
Lele Fu (Sun Yat-sen University), Chuan Chen (Sun Yat-sen University)
Domain AdaptationFederated LearningContrastive LearningImage
🎯 What it does: The FedLSA method is proposed, which utilizes learnable semantic anchors and hyperspherical contrastive learning to address the domain shift problem in federated learning.
Beyond Graph Convolution: Multimodal Recommendation with Topology-aware MLPs
Junjie Huang (Shanghai Jiao Tong University), Weinan Zhang (Shanghai Jiao Tong University)
Recommendation SystemGraph Neural NetworkTextMultimodality
🎯 What it does: A topology-aware multimodal recommendation framework TMLP is proposed, using MLP instead of GCN to capture complex multimodal relationships and avoid over-smoothing.
Beyond Homophily: Graph Contrastive Learning with Macro-Micro Message Passing
Yiyuan Chen (Nanjing University of Aeronautics and Astronautics), Tianzi Zang (Nanjing University of Aeronautics and Astronautics)
Representation LearningGraph Neural NetworkContrastive LearningGraph
🎯 What it does: This paper proposes a graph contrastive learning framework (M³P‑GCL) that simultaneously utilizes structural views and attribute views, enhancing the representation capability for different homogeneous graphs through macro-level view fusion and micro-level adaptive self-propagation.
Beyond Human Data: Aligning Multimodal Large Language Models by Iterative Self-Evolution
Wentao Tan (South China University of Technology), Changxing Ding (South China University of Technology)
RecognitionGenerationTransformerLarge Language ModelReinforcement LearningVision Language ModelImageMultimodality
🎯 What it does: A multimodal self-evolution framework (SENA) is constructed using unlabeled images through three main mechanisms: self-questioning, self-enhanced answering, and image content alignment, achieving continuous improvement of LLM in visual tasks without the need for manual or external model annotations.
Beyond IID: Optimizing Instruction Finetuning from the Perspective of Instruction Interaction and Dependency
Hanyu Zhao (Beijing Academy of Artificial Intelligence), Tengfei Pan (Beijing Academy of Artificial Intelligence)
OptimizationTransformerLarge Language ModelSupervised Fine-TuningText
🎯 What it does: This paper proposes a systematic analysis and optimization of the instruction set of large language models (SFT) from the perspective of instruction interaction and dependency.
Beyond Mandatory Federations: Balancing Egoism, Utilitarianism and Egalitarianism in Mixed-Motive Games
Shaokang Dong (Nanjing Normal University), Yang Gao (Nanjing University)
Federated LearningReinforcement LearningSequential
🎯 What it does: Proposed a Federated Participation Framework (FPF) and Local Multi-Federation (LMF) framework to address the issues of low information sharing and participation in traditional mandatory federations; simultaneously balance the three main objectives of self-interest, utilitarianism, and equality in mixed-motivation games; experiments demonstrate that FPF/LMF outperforms various baselines and reduces the self-interest of non-participants.
Beyond Monotonicity: On the Convergence of Learning Algorithms in Standard Auction Games
Martin Bichler (Technical University of Munich), Barbara Wohlmuth (Technical University of Munich)
🎯 What it does: This paper studies the Bayes-Nash equilibrium problem of first-price and second-price sealed-bid auctions under the symmetric independent private value model, constructs an infinite-dimensional variational inequality framework, and proves that even without satisfying monotonicity or Minty conditions, the auction game still has a unique VI solution, and the gradient learning algorithm will converge to this equilibrium.
Beyond Pixel and Object: Part Feature as Reference for Few-Shot Video Object Segmentation
Naisong Luo (University of Science and Technology of China), Tianzhu Zhang (University of Science and Technology of China)
SegmentationConvolutional Neural NetworkVideo
🎯 What it does: An end-to-end Part Agent Learning Network (PALN) is designed to achieve few-shot video object segmentation by learning part features.
Beyond Prompt Engineering: A Reinforced Token-Level Input Refinement for Large Language Models
Guang Huang (University of Macau), Pengyang Wang (University of Macau)
OptimizationTransformerLarge Language ModelReinforcement LearningPrompt EngineeringText
🎯 What it does: A token-level input optimization framework based on reinforcement learning, RTLIR, is proposed to automatically remove irrelevant information from LLM inputs.
Beyond Single Emotion: Multi-label Approach to Conversational Emotion Recognition
Yujin Kang (Chung Ang University), Yoon-Sik Cho (Chung Ang University)
RecognitionTransformerContrastive LearningText
🎯 What it does: A multi-label emotion recognition framework ML-ERC is proposed, transforming single-label emotion recognition into a multi-label task, and enhancing model performance through pseudo multi-label annotation and weighted supervised contrastive learning.
Beyond Skip Connection: Pooling and Unpooling Design for Elimination Singularities
Chengkun Sun (University of Florida), Jie Xu (University of Florida)
ClassificationSegmentationConvolutional Neural NetworkTransformerImageBiomedical Data
🎯 What it does: Proposed and implemented the Pool Skip module, which combines max pooling, max unpooling, 3×3 convolution, and skip connections to alleviate the learning degradation caused by the elimination of singularities in deep CNNs.
Beyond Spatial Domain: Cross-domain Promoted Fourier Convolution Helps Single Image Dehazing
Xiaozhe Zhang (Beihang University), Haopeng Zhang (Beihang University)
RestorationConvolutional Neural NetworkMixture of ExpertsImage
🎯 What it does: A joint spatial and frequency domain convolution single image dehazing network (JSFC-Net) is proposed.
Beyond Text: Fine-Grained Multi-Modal Fact Verification with Hypergraph Transformers
Hui Pang (Beijing University of Posts and Telecommunications), Xi Zhang (Beihang University)
RetrievalGraph Neural NetworkTransformerImageTextMultimodality
🎯 What it does: A multi-modal fact-checking framework based on hypergraph Transformer, HGTMFC, is proposed, which utilizes hypergraphs to capture higher-order associations between text and image evidence and enhances information propagation through line graphs.
BeyondGender: A Multifaceted Bilingual Dataset for Practical Sexism Detection
Xuan Luo (Harbin Institute of Technology), Ruifeng Xu (Harbin Institute of Technology)
ClassificationTransformerLarge Language ModelSupervised Fine-TuningPrompt EngineeringText
🎯 What it does: A large-scale bilingual (English-Chinese) gender discrimination detection dataset called BeyondGender has been created, with six fine-grained labels (gender, wording, hate, malice, etc.) designed to distinguish different forms of gender discrimination and hatred towards men.
BGDB: Bernoulli-Gaussian Decision Block with Improved Denoising Diffusion Probabilistic Models
Chengkun Sun (University of Florida), Jie Xu (University of Florida)
ClassificationRestorationSegmentationDiffusion modelImage
🎯 What it does: A module based on Bernoulli-Gaussian Decision Blocks (BGDB) is proposed, which simulates the probability distribution of multiple Bernoulli trials using an improved Denoising Diffusion Probabilistic Model (IDDPM) in a single training session, thereby enhancing the stability and performance of the discriminative classifier.
BGHR: Bridging the Gap Between HBox-Supervised and RBox-Supervised Oriented Object Detection via Adaptive Fine-Grained Sample Mining
Chenlin Fu (Shenzhen University), Yingying Zhu (Shenzhen University)
Object DetectionImage
🎯 What it does: A rotation target detection framework BGHR based on horizontal box annotations has been developed, utilizing adaptive fine-grained sample mining and self-supervised branch loss to enhance detection performance.
Bi-Directional Multi-Scale Graph Dataset Condensation via Information Bottleneck
Xingcheng Fu (Guangxi Normal University), Xianxian Li (Beihang University)
Knowledge DistillationGraph Neural NetworkContrastive LearningGraph
🎯 What it does: This paper proposes the BiMSGC framework, which first selects the optimal medium-scale subgraph through information bottleneck, and then performs bidirectional (from large to small and from small to large) graph dataset distillation based on this, ultimately generating multi-scale high-quality synthetic graphs.
Bi-level Contrastive Learning for Knowledge-Enhanced Molecule Representations
Pengcheng Jiang (University of Illinois Urbana Champaign), Jiawei Han (University of Illinois Urbana Champaign)
Representation LearningDrug DiscoveryGraph Neural NetworkContrastive LearningGraph
🎯 What it does: This paper proposes a molecular representation framework called GODE based on dual-layer contrastive learning, which jointly pre-trains molecular graphs and subgraph knowledge graphs centered on the molecule (MolKG) to enhance molecular property prediction performance.
Bi-Level Optimization for Semi-Supervised Learning with Pseudo-Labeling
Marzi Heidari (Carleton University), Yuhong Guo (Carleton University)
ClassificationOptimizationConvolutional Neural NetworkContrastive LearningImage
🎯 What it does: A dual-layer optimization-based pseudo-label learning framework (BOPL) is proposed, which directly optimizes pseudo-labels in the upper layer and optimizes model parameters in the lower layer, thereby enhancing the effectiveness of semi-supervised learning.
Biased Incomplete Multi-View Learning
Haishun Chen (Xidian University), Jinlong Liu (Xidian University)
ClassificationOptimizationMultimodality
🎯 What it does: A reliable incomplete multi-view learning method RIML is proposed to address the issue of view missing pattern bias. It utilizes a cross-category association matrix for distribution calibration and employs distribution sampling that is independent of learning to fill in the missing views, while also introducing a category-aware enhanced focal loss to improve classification performance.
BiDeV: Bilateral Defusing Verification for Complex Claim Fact-Checking
Yuxuan Liu (Renmin University of China), Rui Yan (Renmin University of China)
TransformerLarge Language ModelTextRetrieval-Augmented Generation
🎯 What it does: Designed and implemented the BiDeV framework, achieving de-ambiguation and de-redundancy of complex claims through multi-role collaboration of LLMs, thereby enabling more accurate fact-checking.
Bidirectional Logits Tree: Pursuing Granularity Reconcilement in Fine-Grained Classification
Zhiguang Lu (Chinese Academy of Sciences), Qingming Huang (Chinese Academy of Sciences)
ClassificationTransformerSupervised Fine-TuningImage
🎯 What it does: This paper proposes two methods: Bidirectional Logits Tree (BiLT) and Adaptive Intra-Granularity Difference Learning (AIGDL) to address the feature learning bias caused by the competition between coarse and fine levels in fine-grained classification.
BIG-FUSION: Brain-Inspired Global-Local Context Fusion Framework for Multimodal Emotion Recognition in Conversations
Yusong Wang (Guangdong Institute of Intelligence Science and Technology), Mingkun Xu (Dalian University of Technology)
RecognitionSpiking Neural NetworkTransformerContrastive LearningMultimodality
🎯 What it does: Proposes the BIG-FUSION framework, which combines dual attention Transformer and dual evaluation graph enhancement, and introduces Spiking Neural Networks (SNN) to accomplish multimodal emotion recognition tasks;
BigMac: A Communication-Efficient Mixture-of-Experts Model Structure for Fast Training and Inference
Zewen Jin (University of Science and Technology of China), Cheng Li (Huawei Technologies)
Mixture of ExpertsText
🎯 What it does: A Mixture-of-Experts structure named BigMac is proposed, which significantly reduces All-to-All communication overhead by inserting dimensionality reduction and expansion projections before and after communication between experts, while maintaining the advantages of small experts.
BiMAC: Bidirectional Multimodal Alignment in Contrastive Learning
Masoumeh Zareapoor (Shanghai Jiaotong University), Yue Lu (East China Normal University)
GenerationRetrievalTransformerContrastive LearningImageTextMultimodality
🎯 What it does: A bidirectional multimodal alignment framework BiMAC is proposed, which integrates contrastive learning, image-text generation, and image reconstruction to achieve bidirectional information interaction between text and images.
BindGPT: A Scalable Framework for 3D Molecular Design via Language Modeling and Reinforcement Learning
Artem Zholus (Mila - Quebec AI Institute), Alex Zhavoronkov (Insilico Medicine AI Limited)
Drug DiscoveryTransformerLarge Language ModelReinforcement LearningTextGraph
🎯 What it does: A GPT-based BindGPT framework is proposed, which generates 3D molecules and their conformations in protein binding pockets using text representations (SMILES+XYZ) without prior invariant assumptions.
Bites of Tomorrow: Personalized Recommendations for a Healthier and Greener Plate
Jiazheng Jing (Nanyang Technological University), Chunyan Miao (Nanyang Technological University)
Recommendation SystemTransformerTabularAgriculture Related
🎯 What it does: Proposes a green food recommendation task and designs the GRAPE model to achieve dual optimization of personalization and sustainability.
Black-Box Test-Time Prompt Tuning for Vision-Language Models
Fan'an Meng (Shandong University of Finance and Economics), Shuai Gong (Shandong University)
Domain AdaptationOptimizationTransformerPrompt EngineeringVision Language ModelImage
🎯 What it does: This paper proposes a black-box testing prompt tuning framework (B₂TPT) to achieve VLM adaptation in a gradient-free environment.
BLADE: Enhancing Black-Box Large Language Models with Small Domain-Specific Models
Haitao Li (Tsinghua University), Yiqun Liu (Tsinghua University)
TransformerLarge Language ModelPrompt EngineeringText
🎯 What it does: By constructing a small domain-specific language model and working in conjunction with a general large language model, the question-and-answer performance in vertical fields (such as law and medicine) is enhanced.
Blend the Separated: Mixture of Synergistic Experts for Data-Scarcity Drug-Target Interaction Prediction
Xinlong Zhai (Beijing University of Posts and Telecommunications), Chuan Shi (Beijing University of Posts and Telecommunications)
Drug DiscoveryGraph Neural NetworkMixture of ExpertsBiomedical Data
🎯 What it does: A hybrid expert model named MoseDTI is proposed to simultaneously handle structural information and knowledge graph information in drug-target interaction prediction, alleviating issues of missing input data and sparse annotations through a self-supervised pseudo-labeling approach.
Block-Based Multi-Scale Image Rescaling
Jian Li (Hunan University), Siwang Zhou (Hunan University)
RestorationSuper ResolutionConvolutional Neural NetworkImage
🎯 What it does: Proposes the Block-Based Multi-Scale Image Rescaling (BBMR) framework, which adaptively allocates scaling rates to image sub-blocks based on SR difficulty at the Downscaling end, and uses JointSR at the Upscaling end to eliminate block effects, achieving high-quality super-resolution.
BloomScene: Lightweight Structured 3D Gaussian Splatting for Crossmodal Scene Generation
Xiaolu Hou (Fudan University), Lihua Zhang (Fudan University)
GenerationData SynthesisDepth EstimationCompressionDiffusion modelGaussian SplattingTextMultimodality
🎯 What it does: We propose BloomScene, a lightweight and structured 3D Gaussian splatting framework that can automatically generate high-quality 3D scenes from cross-modal inputs such as text or images.
BLS-GAN: A Deep Layer Separation Framework for Eliminating Bone Overlap in Conventional Radiographs
Haolin Wang (Hokkaido University), Tamotsu Kamishima (Hokkaido University)
RestorationSegmentationGenerationGenerative Adversarial NetworkImage
🎯 What it does: This paper proposes a BLS-GAN framework that can separate the upper and lower bone layers from a single hand X-ray image, eliminating bone overlap.
BOIDS: High-Dimensional Bayesian Optimization via Incumbent-Guided Direction Lines and Subspace Embeddings
Lam Ngo (RMIT University), Hongyu Zhang (Chongqing University)
OptimizationHyperparameter SearchReinforcement LearningTabular
🎯 What it does: A new high-dimensional Bayesian optimization method called BOIDS is proposed, which enhances the search efficiency on high-dimensional expensive black-box functions by utilizing direction lines guided by dominant solutions and subspace embedding.
Boosting Causal Structure Learning: An Asymmetric Exponential Modulation Gaussian-Based Adaptive Sample Reweighting Framework
Wei Xiao (Harbin Engineering University), Nianbin Wang (Harbin Engineering University)
Graph Neural NetworkGaussian SplattingGraph
🎯 What it does: This paper proposes an adaptive sample reweighting framework based on an asymmetric exponential modified Gaussian function (DAG-AEG) to enhance the effectiveness of differentiable score-based causal structure learning.
Boosting Consistency in Story Visualization with Rich-Contextual Conditional Diffusion Models
Fei Shen (Nanjing University of Science and Technology), Yang Wei
GenerationData SynthesisTransformerDiffusion modelImageMultimodality
🎯 What it does: A two-stage Rich-Contextual Conditional Diffusion Model (RCDM) framework is proposed to achieve semantic and temporal consistency in story visualization generation.
Boosting Fine-Grained Visual Anomaly Detection with Coarse-Knowledge-Aware Adversarial Learning
Qingqing Fang (Sun Yat-sen University), Jianxing Yu (Sun Yat-sen University)
Anomaly DetectionConvolutional Neural NetworkAuto EncoderGenerative Adversarial NetworkImageBiomedical DataMagnetic Resonance ImagingComputed Tomography
🎯 What it does: By utilizing a small number of coarsely labeled anomalous samples, an energy discriminator drives an autoencoder to align the reconstruction results with normal samples in the feature space, thereby improving the accuracy of fine-grained visual anomaly detection and localization.
Boosting Image De-Raining via Central-Surrounding Synergistic Convolution
Long Peng (University of Science and Technology of China), Zheng-Jun Zha (University of Science and Technology of China)
RestorationConvolutional Neural NetworkImage
🎯 What it does: A center-periphery collaborative convolution (SC) is designed, which replaces conventional convolution (VC) without changing the original network structure, thereby enhancing the performance of various single-image rain and dirt removal models.
Boosting Multimodal Large Language Models with Visual Tokens Withdrawal for Rapid Inference
Zhihang Lin (Xiamen University), Rongrong Ji (Xiamen University)
RecognitionComputational EfficiencyTransformerLarge Language ModelVision Language ModelImageVideoTextMultimodality
🎯 What it does: The Visual Tokens Withdrawal (VTW) module is proposed, which can remove visual tokens in deep networks during inference of multimodal large language models (MLLM), significantly reducing computational overhead.
Boosting Segment Anything Model Towards Open-Vocabulary Learning
Xumeng Han (University of Chinese Academy of Sciences), Qi Tian (Huawei Inc.)
Object DetectionSegmentationTransformerVision Language ModelContrastive LearningImageText
🎯 What it does: This paper presents Sambor, an end-to-end open-source vocabulary object detector that integrates the Segment Anything Model (SAM) with the visual language model (CLIP) to achieve the localization and recognition of objects of any category.
Boosting Short Text Classification with Multi-Source Information Exploration and Dual-Level Contrastive Learning
Yonghao Liu (Jilin University), Renchu Guan (Jilin University)
ClassificationGraph Neural NetworkContrastive LearningText
🎯 What it does: The MI-DELIGHT model is proposed, which significantly improves short text classification performance through multi-source information exploration (words, POS, entity graphs) and dual-layer contrastive learning (instance-level ICL and cluster-level CCL), employing a hierarchical task architecture.
Boosting Test Performance with Importance Sampling--a Subpopulation Perspective
Hongyu Shen (University of Illinois), Zhizhen Zhao (University of Illinois)
ClassificationDomain AdaptationImageText
🎯 What it does: A subpopulation bias analysis framework (DBA) based on importance sampling is proposed, and within this framework, a subpopulation bias correction method (DBCM) for a single estimator is designed to reweight the training set to improve testing performance under subpopulation imbalance.
Boosting Vision State Space Model with Fractal Scanning
Haoke Xiao (Vivo Mobile Communication Company), Bo Li (Vivo Mobile Communication Company)
ClassificationObject DetectionSegmentationConvolutional Neural NetworkImage
🎯 What it does: This paper presents FractalMamba, which utilizes fractal scanning curves for image serialization, enhancing the performance of SSM in visual tasks.
Boosting ViT-based MRI Reconstruction from the Perspectives of Frequency Modulation, Spatial Purification, and Scale Diversification
Yucong Meng (Fudan University), Zhijian Song (Fudan University)
RestorationTransformerMixture of ExpertsBiomedical DataMagnetic Resonance Imaging
🎯 What it does: Proposes the FPS-Former framework, which enhances the performance of ViT in MRI reconstruction through three main modules: frequency modulation, spatial purification, and scale diversification.
Bootstraping Clustering of Gaussians for View-consistent 3D Scene Understanding
Wenbo Zhang (Dalian University of Technology), Huchuan Lu (University of Electronic Science and Technology of China)
Object DetectionSegmentationContrastive LearningGaussian SplattingPoint Cloud
🎯 What it does: This paper presents FreeGS, an unsupervised 3D Gaussian Splatting framework that utilizes the IDentity-coupled Semantic Field (IDSF) to simultaneously encode semantic features and cross-view instance indexing on Gaussians, achieving view-consistent 3D scene understanding.
Bootstrapped Reward Shaping
Jacob Adamczyk (University of Massachusetts Boston), Rahul V. Kulkarni (Texas Tech University)
Reinforcement LearningTabular
🎯 What it does: This paper proposes a bootstrapping reward shaping method (BSRS) that dynamically shapes rewards by using the agent's current estimated state value function as a potential function.
Bootstrapping Heterogeneous Graph Representation Learning via Large Language Models: A Generalized Approach
Hang Gao (National Key Laboratory of Space Integrated Information System Institute of Software Chinese Academy of Sciences), Huaping Liu (Tsinghua University)
Representation LearningGraph Neural NetworkLarge Language ModelGraph
🎯 What it does: This paper proposes a general heterogeneous graph representation learning framework GHGRL, which utilizes large language models to automatically identify the types and formats of nodes and edges, and combines it with a GNN with adaptive parameters for graph information aggregation, without the need for pre-provided type labels or unified feature formats.
Both Supply and Precision: Sample Debias and Ranking Consistency Joint Learning for Large Scale Pre-Ranking System
Feng Gao (Sina Weibo Inc.), Jie Liu (Sina Weibo Inc.)
Recommendation SystemKnowledge DistillationContrastive LearningTabular
🎯 What it does: A pre-sorting framework SDCL is proposed to jointly address the issues of sample selection bias and ranking consistency;
BotSim: LLM-Powered Malicious Social Botnet Simulation
Boyu Qiao (Institute of Information Engineering, Chinese Academy of Sciences), Songlin Hu (Institute of Information Engineering, Chinese Academy of Sciences)
Anomaly DetectionGraph Neural NetworkLarge Language ModelPrompt EngineeringTextGraph
🎯 What it does: This paper designs a scalable LLM-driven malicious social bot simulation framework called BotSim, and based on this framework, generates the BotSim-24 dataset, which includes real human users and LLM-driven bot accounts.
Boundary Decomposition for Finding Nadir Objective Vector in Multi-Objective Discrete Optimization
Ruihao Zheng (Southern University of Science and Technology), Zhenkun Wang (Southern University of Science and Technology)
Optimization
🎯 What it does: A boundary decomposition-based bi-level optimization algorithm BDNC is proposed to accurately calculate the nadir vector of multi-objective discrete optimization problems (MODOP).
Boundary-Aware Temporal Dynamic Pseudo-Supervision Pairs Generation for Zero-Shot Natural Language Video Localization
Xiongwen Deng (Xi'an Jiaotong University), Jihua Zhu (Fuzhou University)
SegmentationRetrievalTransformerLarge Language ModelVision Language ModelVideoText
🎯 What it does: This method proposes a pseudo-supervised framework for zero-shot natural language video localization based on large language models, which can automatically generate time-query pairs for the original video without manual annotation.
Bounded Rationality Equilibrium Learning in Mean Field Games
Yannick Eich (Technische Universitat Darmstadt), Heinz Koeppl (Technische Universitat Darmstadt)
Reinforcement Learning
🎯 What it does: This paper studies the introduction of bounded rationality and receding horizon methods in Mean Field Games (MFG), proposing Quantal Response Equilibria (QRE) and its variant under limited foresight, and designing general fixed point iteration and fictitious game algorithms to learn these equilibria.
Braess’s Paradox of Generative AI
Boaz Taitler (Technion Israel Institute of Technology), Omer Ben-Porat (Technion Israel Institute of Technology)
OptimizationLarge Language Model
🎯 What it does: A dynamic model is proposed and analyzed regarding the competition between generative AI and traditional human Q&A platforms, studying its profit maximization and long-term social impact.
BrainGuard: Privacy-Preserving Multisubject Image Reconstructions from Brain Activities
Zhibo Tian (Lanzhou University), Yi Yang (Zhejiang University)
Data SynthesisFederated LearningSafty and PrivacyDiffusion modelContrastive LearningImageBiomedical DataMagnetic Resonance Imaging
🎯 What it does: The BRAINGUARD framework is proposed, utilizing multi-agent fMRI collaborative training to achieve image reconstruction under privacy protection;
BrainMAP: Learning Multiple Activation Pathways in Brain Networks
Song Wang (University of Virginia), Jundong Li (University of Virginia)
OptimizationExplainability and InterpretabilityGraph Neural NetworkMixture of ExpertsContrastive LearningGraphBiomedical DataMagnetic Resonance Imaging
🎯 What it does: This paper proposes the BrainMAP framework for learning and interpreting multiple activation pathways in fMRI functional connectivity maps, enhancing brain network prediction and interpretability.
Breaking Barriers in Physical-World Adversarial Examples: Improving Robustness and Transferability via Robust Feature
Yichen Wang (Huazhong University of Science and Technology), Minghui Li (Huazhong University of Science and Technology)
Adversarial AttackVision Language ModelImage
🎯 What it does: A robust feature coverage-based physical world adversarial attack method (RFCoA) is proposed, which generates adversarial samples through robust feature injection and semantic pattern minimization.
Breaking Barriers: A Paradigm Shift in Technology Accessibility for Individuals with Physical Disabilities
Kshitij Mishra (Indian Institute of Technology Patna), Asif Ekbal (Indian Institute of Technology Jodhpur)
Reinforcement LearningText
🎯 What it does: This paper presents EDiSS, an empathetic support dialogue system for individuals with physical disabilities.
Breaking Data Silos in Parkinson’s Disease Diagnosis: An Adaptive Federated Learning Approach for Privacy-Preserving Facial Expression Analysis
Meng Pang (Nanchang University), Binghui Wang (Illinois Institute of Technology)
Federated LearningSafty and PrivacyConvolutional Neural NetworkTransformerImageBiomedical Data
🎯 What it does: An adaptive federated learning framework based on client contribution assessment is proposed, which collaboratively trains on multi-institutional PD facial expression data to address data silos and privacy issues.
Breaking Information Isolation: Accelerating MRI via Inter-sequence Mapping and Progressive Masking
Jianwei Zheng (Zhejiang University of Technology), Jiawei Jiang (Zhejiang University of Technology)
RestorationOptimizationAuto EncoderImageBiomedical DataMagnetic Resonance Imaging
🎯 What it does: A new MRI acceleration method called Information Coupling MRI Acceleration (IMA) is proposed to address the issue of information isolation in multi-sequence MRI reconstruction.
Breaking Symmetries in Quantified Graph Search: A Comparative Study
Mikoláš Janota (Czech Technical University in Prague), Stefan Szeider (Algorithms and Complexity Group TU Wien)
Graph
🎯 What it does: This paper studies how to utilize symmetry elimination techniques in quantified graph search to avoid the generation of redundant graphs, and extends the SAT Modulo Symmetries (SMS) framework to Quantified Boolean Formulas (QBF);
Bridge 2D-3D: Uncertainty-aware Hierarchical Registration Network with Domain Alignment
Zhixin Cheng (University of Science and Technology of China), Tianzhu Zhang (University of Science and Technology of China)
Pose EstimationDomain AdaptationConvolutional Neural NetworkTransformerImagePoint Cloud
🎯 What it does: A hierarchical registration network based on uncertainty perception is designed and implemented, capable of cross-modal registration between images and point clouds, and enhancing robustness through domain alignment.
Bridge Diffusion Model: Bridge Chinese Text-to-Image Diffusion Model with English Communities
Shanyuan Liu (360 AI Research), Yuhui Yin (360 AI Research)
GenerationData SynthesisDiffusion modelImageText
🎯 What it does: Proposes the Bridge Diffusion Model (BDM), which generates images from Chinese text through a backend-branch architecture, while also being compatible with English community plugins.
Bridge Then Begin Anew: Generating Target-Relevant Intermediate Model for Source-Free Visual Emotion Adaptation
Jiankun Zhu (Harbin Institute of Technology), Hongxun Yao (Harbin Institute of Technology)
RecognitionDomain AdaptationKnowledge DistillationImage
🎯 What it does: A two-stage framework for source-agnostic visual emotion recognition is proposed, consisting of BBA (Domain Bridge Model Generation + Target-Related Model Adaptation).
Bridging Knowledge Gap Between Image Inpainting and Large-Area Visible Watermark Removal
Yicheng Leng (Xidian University), Guanbin Li (Sun Yat-sen University)
RestorationGenerative Adversarial NetworkImage
🎯 What it does: A framework for visible watermark removal based on image inpainting is proposed, utilizing residual background information to enhance the removal effect of large-area watermarks.
Bridging Molecular Graphs and Large Language Models
Runze Wang (Dalian University of Technology), Yanming Shen (Dalian University of Technology)
Drug DiscoveryGraph Neural NetworkTransformerLarge Language ModelPrompt EngineeringTextMultimodalityGraph
🎯 What it does: The molecular graph structure is mapped to a special graph token through a graph encoder, and this token is aligned to the LLM's vocabulary space using a cross-attention mechanism, allowing the LLM to understand and reason about the molecular graph while maintaining the original weights; subsequently, the IUPAC name and task description are added to the prompt, and the task head is used to output the prediction results.
Bridging Sequence-Structure Alignment in RNA Foundation Models
Heng Yang (University of Exeter), Ke Li (University of Exeter)
TransformerLarge Language ModelSequentialBiomedical Data
🎯 What it does: The OmniGenome RNA base model is proposed, achieving bidirectional alignment of sequences and secondary structures, and performing excellently in RNA design and structure prediction tasks.
Bridging the Gap Between Hyperdimensional Computing and Kernel Methods via the Nyström Method
Quanling Zhao (University of California San Diego), Tajana Rosing (University of California San Diego)
ClassificationRepresentation LearningGraph Neural NetworkTextGraph
🎯 What it does: A Hyperdimensional Computing (HDC) encoding algorithm called NysHD based on the Nyström method is proposed, which can map any positive definite kernel function to low-precision high-dimensional vectors, achieving efficient HDC learning.
Bridging the Gap for Test-Time Multimodal Sentiment Analysis
Zirun Guo (Zhejiang University), Yangyang Wu (Zhejiang University)
Domain AdaptationContrastive LearningVideoTextMultimodalityAudio
🎯 What it does: This paper proposes a test-time adaptive method for multimodal emotion analysis (CASP), which addresses the issue of model performance degradation under target domain distribution shift by randomly dropping multimodal inputs and ensuring consistency through contrastive learning, while generating stable pseudo-labels for self-supervised training.
Bridging the Semantic Granularity Gap Between Text and Frame Representations for Partially Relevant Video Retrieval
WooJin Jun (Sungkyunkwan University), Jae-Pil Heo (Sungkyunkwan University)
RetrievalTransformerContrastive LearningVideoText
🎯 What it does: This paper studies a new frame-level retrieval framework that addresses the issue of semantic granularity mismatch between text and video frames, thereby improving the performance of partial relevant video retrieval (PRVR).
Bridging the User-side Knowledge Gap in Knowledge-aware Recommendations with Large Language Models
Zheng Hu (University of Electronic Science and Technology of China), Fuji Ren (University of Electronic Science and Technology of China)
Recommendation SystemGraph Neural NetworkLarge Language ModelAuto EncoderContrastive LearningGraph
🎯 What it does: This paper utilizes large language models to infer user interests from historical behavior, constructing a Collaborative Interest Knowledge Graph (CIKG) for recommendations.
Bridging Traffic State and Trajectory for Dynamic Road Network and Trajectory Representation Learning
Chengkai Han (Beihang University), Junjie Wu (Beihang University)
Autonomous DrivingRepresentation LearningGraph Neural NetworkTransformerContrastive LearningGraphTime Series
🎯 What it does: By jointly modeling traffic state data and trajectory data, the TRACK framework is proposed to learn low-dimensional representations of dynamic road networks and trajectories.
Bridging Training and Execution via Dynamic Directed Graph-Based Communication in Cooperative Multi-Agent Systems
Zhuohui Zhang (Tongji University), Gang Li (Tongji University)
Graph Neural NetworkTransformerReinforcement LearningGraph
🎯 What it does: This paper proposes a Transformer-based Layered Graph Aggregation Network (TGCNet) that utilizes dynamic directed graphs to learn communication strategies among multiple agents. During centralized training, it approximates the global state through a graph aggregation network, and during decentralized execution, it uses a Transformer decoder to complete feature fusion.
Bright-NeRF: Brightening Neural Radiance Field with Color Restoration from Low-Light RAW Images
Min Wang (Northwestern Polytechnical University), Qing Wang (Northwestern Polytechnical University)
RestorationGenerationNeural Radiance FieldImage
🎯 What it does: An unsupervised low-light RAW image NeRF reconstruction method is proposed, capable of generating new perspective images that conform to normal lighting in low-light environments.
BSAFusion: A Bidirectional Stepwise Feature Alignment Network for Unaligned Medical Image Fusion
Huafeng Li (Kunming University of Science and Technology), Yafei Zhang (Kunming University of Science and Technology)
TransformerImageMultimodalityBiomedical DataMagnetic Resonance ImagingComputed TomographyPositron Emission Tomography
🎯 What it does: This paper presents BSAFusion, a single-stage bidirectional step feature alignment network for the registration and fusion of unaligned multimodal medical images.
BSDB-Net: Band-Split Dual-Branch Network with Selective State Spaces Mechanism for Monaural Speech Enhancement
Cunhang Fan (Anhui University), Zhao Lv
RestorationAudio
🎯 What it does: A network based on band splitting and dual branches (BSDB-Net) is designed for monaural speech enhancement, with the dual branches processing the magnitude spectrum and the complex spectrum respectively, and sequence modeling is performed using Mamba.
BUFF: Bayesian Uncertainty Guided Diffusion Probabilistic Model for Single Image Super-Resolution
Zihao He (Xiamen University), Yan Zhang
RestorationSuper ResolutionDiffusion modelImage
🎯 What it does: Designed the BUFF model, introducing Bayesian uncertainty masks into diffusion-based single image super-resolution to enhance detail reconstruction through adaptive noise adjustment.
Building a Multi-modal Spatiotemporal Expert for Zero-shot Action Recognition with CLIP
Yating Yu (Northwestern Polytechnical University), Yanning Zhang (Northwestern Polytechnical University)
RecognitionKnowledge DistillationTransformerLarge Language ModelVision Language ModelVideoTextMultimodality
🎯 What it does: A multi-modal spatio-temporal expert framework STDD based on CLIP is proposed, specifically addressing the problem of collaborative understanding of visual and textual spatio-temporal dynamics in zero-shot action recognition (ZSAR).
C2F-TP: A Coarse-to-Fine Denoising Framework for Uncertainty-Aware Trajectory Prediction
Zichen Wang (Central South University), Jian Zhang (Central South University)
GenerationAutonomous DrivingRecurrent Neural NetworkDiffusion modelTime SeriesSequential
🎯 What it does: C2F-TP is proposed, using a coarse-to-fine denoising framework for vehicle trajectory prediction. It first learns the multimodal trajectory distribution through a spatial-temporal interaction module, and then gradually denoises to generate more accurate trajectories through a conditional denoising module.
C2P-CLIP: Injecting Category Common Prompt in CLIP to Enhance Generalization in Deepfake Detection
Chuangchuang Tan (Beijing Jiaotong University), Yunchao Wei (Beijing Jiaotong University)
Anomaly DetectionTransformerPrompt EngineeringContrastive LearningImage
🎯 What it does: This study investigates the mechanism of CLIP in deepfake detection and proposes a method called C2P-CLIP to enhance detection performance by injecting category-general prompts into the CLIP visual encoder.
C2PD: Continuity-Constrained Pixelwise Deformation for Guided Depth Super-Resolution
Jiahui Kang (Ocean University of China), Zhi Liu (Shandong University)
RestorationDepth EstimationSuper ResolutionConvolutional Neural NetworkImage
🎯 What it does: A guided depth super-resolution method based on continuity-constrained pixel-level deformation, C2PD, is proposed.
C3oT: Generating Shorter Chain-of-Thought Without Compromising Effectiveness
Yu Kang (Beike Inc), Wei Zou (Beike Inc)
CompressionOptimizationTransformerLarge Language ModelSupervised Fine-TuningPrompt EngineeringTextChain-of-Thought
🎯 What it does: The C3oT framework is proposed, which significantly shortens intermediate reasoning steps through Conditional Compressed Chain-of-Thought (CoT) while maintaining the accuracy of the final answer.
CA-Edit: Causality-Aware Condition Adapter for High-Fidelity Local Facial Attribute Editing
Xiaole Xian (Shenzhen University), Linlin Shen (Great Bay University)
Image TranslationGenerationDiffusion modelImageText
🎯 What it does: This paper proposes a local facial attribute editing method based on diffusion models, called CA-Edit, which achieves high-fidelity attribute modification through text-guided local filling.
CA-MLIF: Cross-Attention and Multimodal Low-Rank Interaction Fusion Framework for Tumor Prognostic Prediction
Yajun An (Chongqing College of Finance and Economics), Xipeng Pan (Guilin University of Electronic Technology)
TransformerMultimodalityBiomedical DataComputed Tomography
🎯 What it does: An end-to-end cross-attention and low-rank interaction fusion framework (CA-MLIF) is proposed to integrate pathological images, CT images, gene expression, and clinical information for tumor survival prediction.
CAD-GPT: Synthesising CAD Construction Sequence with Spatial Reasoning-Enhanced Multimodal LLMs
Siyu Wang (Shanghai Jiao Tong University), Jie Yang (University of Minnesota)
GenerationData SynthesisTransformerLarge Language ModelSupervised Fine-TuningVision Language ModelImageTextMultimodality
🎯 What it does: Developed CAD-GPT, a multimodal large model based on LLaVA, capable of automatically generating CAD construction sequences from a single image or text description.
CAGE: Unsupervised Visual Composition and Animation for Controllable Video Generation
Aram Davtyan (University of Bern), Paolo Favaro (LMU Munich)
GenerationData SynthesisTransformerDiffusion modelVideo
🎯 What it does: This paper proposes an unsupervised video generation model called CAGE, which can achieve scene composition and object animation by placing visual tokens in both spatial and temporal dimensions.
CAKE: Category Aware Knowledge Extraction for Open-Vocabulary Object Detection
Shiyuan Ma (Xiamen University), Shengchuan Zhang (Xiamen University)
Object DetectionKnowledge DistillationConvolutional Neural NetworkImage
🎯 What it does: An open-source vocabulary object detection framework CAKE is proposed, which integrates category-aware knowledge distillation and a global RPN based on feature sets, significantly enhancing the detection capability for new categories.
CALF: Aligning LLMs for Time Series Forecasting via Cross-modal Fine-Tuning
Peiyuan Liu (Tsinghua University), Shu-Tao Xia (Tsinghua University)
TransformerLarge Language ModelSupervised Fine-TuningTime Series
🎯 What it does: By constructing the CALF framework and using a cross-modal fine-tuning method, the text representation of the LLM is aligned with the time series input in terms of distribution, features, and output space, achieving long-term and short-term multivariate forecasting.
Calibrated Disambiguation for Partial Multi-label Learning
Zhuoming Li (Southeast University), Zicong Miao (China Telecom Cloud Computing Corporation)
ClassificationObject DetectionTransformerImage
🎯 What it does: In the multi-label learning task with a candidate label set that includes noisy labels, a calibration-based curriculum learning method PML-CD is proposed. It utilizes a transferable calibrator learned from the confidence histogram during the training process to adaptively assign weights to samples and enhances robustness through prototype alignment regularization.
Calibrating Large Language Models with Sample Consistency
Qing Lyu (University of Pennsylvania), Chris Callison-Burch (University of Pennsylvania)
GenerationReinforcement Learning from Human FeedbackTransformerLarge Language ModelReinforcement LearningText
🎯 What it does: This paper estimates the confidence of large language models (LLMs) by sampling multiple generations and calculating their consistency (using three metrics: agreement, entropy, and FSD), thereby achieving post-processing calibration of the model's prediction results.
CALLIC: Content Adaptive Learning for Lossless Image Compression
Daxin Li (Harbin Institute of Technology), Wen Gao (Peking University)
CompressionTransformerImage
🎯 What it does: This paper proposes CALLIC, a content-adaptive lossless image compression method based on the MDL principle and PETL.
CAMH: Advancing Model Hijacking Attack in Machine Learning
Xing He (Zhejiang University), Shouling Ji (Zhejiang University)
OptimizationAdversarial AttackConvolutional Neural NetworkImage
🎯 What it does: This paper proposes and implements Category-Agnostic Model Hijacking (CAMH), an attack method that can repurpose machine learning models for any new task without degrading the performance of the original task.
CAMSIC: Content-aware Masked Image Modeling Transformer for Stereo Image Compression
Xinjie Zhang (SenseTime Research), Jun Zhang (The Chinese University of Hong Kong Institute for AI Industry Research Tsinghua University)
CompressionTransformerAuto EncoderImage
🎯 What it does: A Transformer-based stereo image compression framework called CAMSIC is proposed, which utilizes content-aware mask image modeling to achieve a decoder-free entropy model. Each image is encoded separately, and a dual-view prior is used to achieve more accurate probability distribution predictions.
Can Generative Models Improve Self-Supervised Representation Learning?
Sana Ayromlou (Vector Institute), Arash Afkanpour (Vector Institute)
GenerationRepresentation LearningDiffusion modelGenerative Adversarial NetworkContrastive LearningImage
🎯 What it does: Utilizing instance conditional generative models (such as Stable Diffusion and ICGAN) to generate semantically consistent image enhancements, enriching the views of self-supervised learning and improving representation quality.
Can Large Language Models Derive High-Level Cognition from Low-Level and Fragmented Foundational Information?
Yang Liu (Hunan University), Kai Lu (National University of Defense Technology)
TransformerLarge Language ModelSupervised Fine-TuningPrompt EngineeringText
🎯 What it does: A dataset specifically designed to assess the high-level cognitive (HLC) abilities of large language models (MatchIntel) has been constructed, and three evaluation tasks (MCQ, SCQ, TFQ) have been proposed. The performance of existing LLMs on HLC tasks has been verified through fine-tuning and comparative experiments.