IJCAI 2025 Papers with AI Summaries
International Joint Conference on Artificial Intelligence · 1014 papers
→ IJCAI 2025 papers with code (343)
Each paper below shows an AI-generated one-line summary. Get the full 6-part summary (innovation, method, data, results, limitations) and search all 1014 IJCAI 2025 papers by keyword, author or institution —
free trial on arXivSub.
2D Gaussian Splatting for Outdoor Scene Decomposition and Relighting
Wei Feng (Tianjin University), Nan Li (Tianjin University)
Neural Radiance FieldGaussian SplattingImage
🎯 What it does: Propose an inverse rendering method based on 2D Gaussian splatting, enabling scene decomposition and relighting under outdoor lighting variations and unknown lighting conditions.
A Case for Validation Buffer in Pessimistic Actor-Critic
Michał Nauman (University of Warsaw), Marek Cygan (University of Warsaw)
Robotic IntelligenceReinforcement LearningBenchmark
🎯 What it does: Propose a strategy called Validation-based Pessimism Regulation (VPL) that dynamically adjusts the level of pessimism using a validation buffer, enhancing the sample efficiency and final performance of actor-critic methods by minimizing the approximation error of the critical network target.
A Centrality-based Graph Learning Framework
Jiajun Yu (Zhejiang University), Haishuai Wang (Zhejiang University)
Representation LearningGraph Neural NetworkContrastive LearningGraph
🎯 What it does: Propose a new readout function ACRead (compatible with GNNs) and a GNN-free framework ACGL, which generates graph-level representations by adaptively fusing multiple node centrality measures through an attention mechanism.
A Correlation Manifold Self-Attention Network for EEG Decoding
Chen Hu (Jiangnan University), Ziheng Chen (University of Trento)
ClassificationTransformerBiomedical Data
🎯 What it does: Propose Correlation Attention Network (CorAtt), implementing self-attention on full-rank correlation matrices, and design corresponding transformation layers, aggregation layers, and tangent mapping layers.
A Cross-Modal Densely Guided Knowledge Distillation Based on Modality Rebalancing Strategy for Enhanced Unimodal Emotion Recognition
Shuang Wu, Ziyu Jia (New York University)
RecognitionKnowledge DistillationConvolutional Neural NetworkTransformerMultimodalityBiomedical Data
🎯 What it does: Propose a cross-modal knowledge distillation framework for emotion recognition, first using a multi-modal teacher network (visual + EEG) to learn fused features, then transferring this knowledge to a single-modal student network that uses only visual input;
A Datalog Rewriting Algorithm for Warded Ontologies
Davide Benedetto (Prometheux), Adriano Vlad-Starrabba (Prometheux)
Data SynthesisComputational Efficiency
🎯 What it does: Designed a Datalog rewriting algorithm called WardedRewrite specifically for warded ontologies and implemented the algorithm;
A Dual Stream Visual Tokenizer for LLM Image Generation
Yongqian Li (Wuhan University), Zhifei Li (Mobvoi Innovation Technology Company Limited)
GenerationTransformerLarge Language ModelDiffusion modelAuto EncoderMultimodality
🎯 What it does: Propose a dual-stream visual tokenizer that combines high-level semantic tokens with low-level pixel tokens to efficiently convert images into sequences for LLMs.
A Dynamic Knowledge Update-Driven Model with Large Language Models for Fake News Detection
Di Jin (Tianjin University), Dongxiao He (Tianjin University)
ClassificationExplainability and InterpretabilityGraph Neural NetworkTransformerLarge Language ModelTextRetrieval-Augmented Generation
🎯 What it does: Proposed a fake news detection framework (DYNAMO) that combines dynamic knowledge graph updates with large language models (LLMs). The framework uses Monte Carlo Tree Search (MCTS) to decompose news into sub-problems, gradually verifying the authenticity of each sub-problem, and ultimately achieving fake news detection while extracting and updating factual knowledge from real news.
A Dynamic Stiefel Graph Neural Network for Efficient Spatio-Temporal Time Series Forecasting
Jiankai Zheng (Wuhan University of Technology), Liang Xie (Wuhan University of Technology)
Autonomous DrivingGraph Neural NetworkTime SeriesFinance Related
🎯 What it does: Propose a Dynamic Stiefel Graph Neural Network (DST-SGNN) for efficient prediction of spatiotemporal sequences
A Fast and Accurate ANN-SNN Conversion Algorithm with Negative Spikes
Xu Wang (Shanghai Institute of Microsystem and Information Technology Chinese Academy of Sciences), Jiamao Li (Shanghai Institute of Microsystem and Information Technology Chinese Academy of Sciences)
OptimizationComputational EfficiencyConvolutional Neural NetworkSpiking Neural NetworkImage
🎯 What it does: Designed a fast and accurate algorithm for converting Artificial Neural Networks (ANN) to Spiking Neural Networks (SNN), supporting negative pulses and applicable to Leaky ReLU networks;
A Fast Neural Architecture Search Method for Multi-Modal Classification via Knowledge Sharing
Zhihua Cui (Taiyuan University of Science and Technology), Zhixia Zhang (Taiyuan University of Science and Technology)
ClassificationNeural Architecture SearchMultimodality
🎯 What it does: Propose a knowledge-sharing based neural architecture search method, KS-NAS, specifically designed for multi-modal classification tasks, which accelerates training and enhances the final model performance by sharing a dynamic knowledge base during the evolutionary search process.
A Fast-Adaptive Cognitive Diagnosis Framework for Computerized Adaptive Testing Systems
Yuanhao Liu (East China Normal University), Aimin Zhou (East China Normal University)
Computational EfficiencyRecurrent Neural NetworkGraph Neural NetworkTransformerTabularSequential
🎯 What it does: Proposes the FACD framework to achieve fast and accurate cognitive diagnosis in computerized adaptive testing (CAT), enhancing early-stage diagnostic quality through dynamic collaboration and personalized diagnostic modules.
A Fine-Grained Complexity View on Propositional Abduction - Algorithms and Lower Bounds
Victor Lagerkvist (Linkoping University), Johannes Schmidt (Jonkping University)
OptimizationComputational Efficiency
🎯 What it does: Investigate the fine-grained complexity of propositional abduction, analyzing enumerability, upper and lower bounds under different constraint languages, particularly focusing on algorithm performance with the number of variables n as a parameter.
A Finite-State Controller Based Offline Solver for Deterministic POMDPs
Alex Schutz (University of Oxford), Nick Hawes (University of Oxford)
OptimizationReinforcement Learning
🎯 What it does: Designed an offline solver called DetMCVI based on a finite state controller for solving deterministic partially observable Markov decision processes (DetPOMDP).
A First Runtime Analysis of NSGA-III on a Many-Objective Multimodal Problem: Provable Exponential Speedup via Stochastic Population Update
Andre Opris (University of Passau)
OptimizationComputational EfficiencyBenchmark
🎯 What it does: This paper rigorously analyzes the runtime of NSGA-III on the multi-objective Jump-Zero-Jump (d-OJZJ) benchmark, providing upper and lower bounds for two variants: without random updates and with random population updates, and proving that random updates can achieve exponential speedup.
A Game-Theoretic Perspective on Inconsistency Handling
Yakoub Salhi (University of Artois)
🎯 What it does: Proposed a game theory framework for restoring consistency through interactive dialogue in a propositional basis, and defined a new inconsistency measure I_w^mc based on this framework.
A General Framework for Representing Controlled Natural Language Sentences and Translation to KR Formalisms
Simone Caruso (University of Genoa), Alice Tarzariol (University of Klagenfurt)
Representation LearningText
🎯 What it does: This paper proposes and implements CNLWizard, a generic framework that can quickly generate the syntax of control natural language (CNL) and corresponding conversion functions through YAML configuration, achieving the translation of CNL sentences into knowledge representation and reasoning (KR) formalisms such as ASP, CP, and SMT.
A Generalized Diffusion Framework with Learnable Propagation Dynamics for Source Localization
Dongpeng Hou (Northwestern Polytechnical University), Xianghua Li (Northwestern Polytechnical University)
Graph Neural NetworkDiffusion modelScore-based ModelGraph
🎯 What it does: Propose a general diffusion framework GDFSL for source localization under different propagation dynamics;
A Hybrid Multi-Factor Network with Dynamic Sequence Modeling for Early Warning of Intraoperative Hypotension
Mingyue Cheng (University of Science and Technology of China), Chunli Liu (Hefei University of Technology)
Anomaly DetectionTransformerTime SeriesBiomedical Data
🎯 What it does: Redefine the early warning of hypotension during surgery as a multivariate time series prediction problem and propose the Hybrid Multi-Factor (HMF) network.
A Little Subsidy Ensures MMS Allocation for Three Agents
Xiaowei Wu (University of Macau), Shengwei Zhou (University of Macau)
Optimization
🎯 What it does: Propose that with only three agents, a total subsidy of 1/6 can achieve a maximin share (MMS) allocation for indivisible items (including goods and chores); and provide a tighter lower bound of 2/49.
A Logic of General Attention Using Edge-Conditioned Event Models
Gaia Belardinelli (Stanford University), Sebastian Watzl (University of Oslo)
Explainability and Interpretability
🎯 What it does: Proposes a theoretical framework for general attention logic, viewing attention as a dynamic modality;
A Logic-Based Approach to Causal Discovery: Signal Temporal Logic Perspective
Nasim Baharisangari (Arizona State University), Zhe Xu (Arizona State University)
Explainability and InterpretabilityDrug DiscoveryTime Series
🎯 What it does: Propose a causal discovery framework based on Signal Temporal Logic (STL), generating interpretable STL causal graphs (STL-CD).
A Logic-based Framework for Decoding Enthymemes in Argument Maps Involving Implicitness in Premises and Claims
Victor David (Université Côte d'Azur), Anthony Hunter (University College London)
OptimizationExplainability and InterpretabilityGraph
🎯 What it does: Proposed a framework based on default logic to explicitly reveal implicit premises and claims in argument graphs, optimizing the decoding of implicit information by maximizing consistency with support/attack labels in the graph.
A Medical Image Classification Network Based on Multi-View Consistent Momentum Contrastive Learning
Chuangui Cao (China University of Mining and Technology), Lili Guo (China University of Mining and Technology)
ClassificationConvolutional Neural NetworkContrastive LearningImageBiomedical Data
🎯 What it does: Proposed a medical image classification model named SVCMC based on tri-perspective consistency momentum contrast learning, combined with Sobel operator for edge enhancement.
A Methodological Framework for Measuring Spatial Labeling Similarity
Yihang Du (Zhongnan University of Economics and Law), Xiaobo Sun (Zhongnan University of Economics and Law)
Representation LearningData-Centric LearningGraph Neural NetworkBiomedical Data
🎯 What it does: Proposed a methodological framework to measure spatial annotation similarity, and implemented specific metrics SLAM in spatial transcriptomics (ST);
A Multi-Granularity Clustering Approach for Federated Backdoor Defense with the Adam Optimizer
Jidong Yuan (Beijing Jiaotong University), Baomin Xu (Beijing Jiaotong University)
OptimizationFederated LearningSafty and PrivacyImageTime Series
🎯 What it does: Proposes a federated learning backdoor defense method named FLAC that combines the Adam optimizer with multi-granularity clustering;
A Multi-view Fusion Approach for Enhancing Speech Signals via Short-time Fractional Fourier Transform
Zikun Jin (Shanxi University), Haijun Geng (Shanxi University)
RestorationConvolutional Neural NetworkAudio
🎯 What it does: A multi-perspective fusion method for single-channel speech enhancement is studied, which constructs multi-perspective inputs using short-time fractional Fourier transform (STFrFT) and STFT together, and achieves speech enhancement through a lightweight UNet combined with multi-scale attention.
A Neuro-Symbolic Framework for Sequence Classification with Relational and Temporal Knowledge
Luca Salvatore Lorello (University of Pisa), Stefano Melacci (University of Siena)
ClassificationConvolutional Neural NetworkImageSequentialBenchmark
🎯 What it does: Propose a multi-stage neuro-symbolic framework for joint utilization of visual perception, constraint reasoning, and temporal reasoning in sequence classification tasks involving relational and temporal knowledge.
A Non-Interventionist Approach to Causal Reasoning Based on Lewisian Counterfactuals
Carlos Aguilera-Ventura, Dmitry Rozplokhas (Tu Wien)
Review/Survey Paper
🎯 What it does: Built a two-dimensional semantics based on Lewis, providing a framework that can express actual causality without intervention, and presented a proof of PSPACE completeness for model checking.
A Novel Local Search Algorithm for the Vertex Bisection Minimization Problem
Rui Sun (Northeast Normal University), Yi Zhou (University of Electronic Science and Technology of China)
OptimizationComputational EfficiencyGraphBenchmark
🎯 What it does: Proposed a new local search algorithm called CELS based on (k,l,S)-cluster for solving the Vertex Bipartition Minimization Problem (VBMP).
A Novel Sparse Active Online Learning Framework for Fast and Accurate Streaming Anomaly Detection Over Data Streams
Zhong Chen (Southern Illinois University Carbondale), Meikang Qiu (Augusta University)
Anomaly DetectionTabularSequential
🎯 What it does: Propose a streaming anomaly detection framework SAD based on sparse active online learning, capable of real-time identification of anomalies in high-dimensional data streams.
A Primal-dual Perspective for Distributed TD-learning
Han Dong Lim, Donghwan Lee (KAIST)
OptimizationReinforcement LearningGraph
🎯 What it does: This paper studies distributed temporal difference learning in networked multi-agent Markov decision processes, proposing an algorithm based on the primal-dual method and providing finite-time convergence analysis.
A Prior-based Discrete Diffusion Model for Social Graph Generation
Shu Yin (Northwestern Polytechnical University), Chao Gao (Northwestern Polytechnical University)
Data SynthesisGraph Neural NetworkDiffusion modelGraph
🎯 What it does: Propose a prior-based discrete diffusion model (PDDM) that generates more realistic social propagation graphs through a discrete forward process and a reverse starting point based on user similarity.
A Priori Estimation of the Approximation, Optimization and Generalization Errors of Random Neural Networks for Solving Partial Differential Equations
Xianliang Xu (Tsinghua University), Zhongyi Huang (Tsinghua University)
OptimizationPhysics Related
🎯 What it does: This paper studies solving partial differential equations using random neural networks (with randomly sampled weights in hidden layers and trainable output weights) within the PINNs framework, and provides a-priori estimates for approximation error, optimization error, and generalization error for Barron-type functions.
A Reduction-Based Algorithm for the Clique Interdiction Problem
Chenghao Zhu (University of Electronic Science and Technology of China), Haoyu Jiang (University of Electronic Science and Technology of China)
OptimizationComputational EfficiencyGraph
🎯 What it does: For the Clique Interdiction Problem, we propose an algorithm called RECIP based on data reduction, and design various new reduction rules in the preprocessing phase.
A SAT-based Method for Counting All Singleton Attractors in Boolean Networks
Rei Higuchi (Kobe University), Naoyuki Tamura
OptimizationComputational EfficiencyGraphBenchmark
🎯 What it does: This paper proposes a counting method based on SAT that can directly count all single-point attractors in a given Boolean network without enumerating individual single-point attractors.
A Sequent Calculus for Answer Set Entailment
Thomas Eiter (Technische Universitaet Wien), Tobias Geibinger (Technische Universitaet Wien)
🎯 What it does: Proposed a sequential reasoning system (ELK) for Equilibrium Logic that guarantees both consistency and completeness
A Simple yet Effective Hypergraph Clustering Network
Qianqian Wang (Xidian University), Quanxue Gao (Xidian University)
Computational EfficiencyRepresentation LearningGraph Neural NetworkGraph
🎯 What it does: Proposed a simple and efficient hypergraph clustering network (HCN), achieving unsupervised clustering through hypergraph smoothing preprocessing, non-shared MLP encoder, Self-Diagonal Consistency module, and Structure Alignment module.
A Structural Complexity Analysis of Hierarchical Task Network Planning
Cornelius Brand (Regensburg University), Simon Wietheger (TU Wien)
Computational Efficiency
🎯 What it does: This paper conducts a fine-grained complexity analysis of plan verification, plan existence, and state reachability problems in Hierarchical Task Networks (HTN), proposes structural metrics that can be polynomially solved in the original network, and provides an algorithm meta-theorem for generalizing to networks containing composite tasks.
A Symmetric Relative-Error Loss Function for Intermittent Multiscale Signal Modelling
Sergio M. Vanegas Arias (LUT University), Fredy Ruiz Palacios (Politecnico di Milano)
OptimizationTime SeriesFinance RelatedPhysics Related
🎯 What it does: This paper proposes two relative error loss functions, Mean Arctangent Squared Percentage Error (MASPE) and Symmetric Mean Arctangent Squared Percentage Error (SMASPE), for regression problems involving multi-scale discontinuous signals.
A Theoretical Perspective on Why Stochastic Population Update Needs an Archive in Evolutionary Multi-objective Optimization
Shengjie Ren (Nanjing University), Chao Qian (Nanjing University)
OptimizationBenchmark
🎯 What it does: This paper theoretically investigates why external archives are needed when using stochastic population update (SPU) in multi-objective evolutionary algorithms (MOEA), and proves that using archives can maintain high-quality solutions while significantly reducing the required population size and improving search efficiency.
A Timestep-Adaptive Frequency-Enhancement Framework for Diffusion-based Image Super-Resolution
Yueying Li (Zhejiang University), Haobo Wang (Zhejiang University)
GenerationSuper ResolutionConvolutional Neural NetworkDiffusion modelImage
🎯 What it does: This paper proposes a Temporal Adaptive Frequency Enhancement Framework (TFDSR) to improve image super-resolution based on diffusion models.
A Unifying Framework for Semiring-Based Constraint Logic Programming With Negation
Jeroen Spaans (Open Universiteit), Jesse Heyninck (Open Universiteit)
🎯 What it does: This paper proposes a unified framework that extends semi-ring based constraint logic programs (SCLP) to support negation as failure, and provides definitions and proofs of its model and revised closure semantics.
A Weighted-Based Fast Local Search for α-Neighbor p-Center Problem
Qingyun Zhang (Huazhong University of Science and Technology), Zhouxing Su (Huazhong University of Science and Technology)
OptimizationGraphBenchmark
🎯 What it does: Propose an algorithm based on Weighted Fast Local Search (WFLS) to solve the α-neighbor p-center problem by transforming the original problem into a series of decision subproblems with given radii and solving them step by step.
A-I-RAVEN and I-RAVEN-Mesh: Two New Benchmarks for Abstract Visual Reasoning
Mikołaj Małkiński (Warsaw University of Technology), Jacek Mańdziuk (Warsaw University of Technology)
Data SynthesisRepresentation LearningImageMeshBenchmark
🎯 What it does: This paper constructs two new Raven Advanced Matrix (RPM) benchmarks—A-I-RAVEN and I-RAVEN-Mesh—to systematically evaluate the generalization and knowledge transfer capabilities of deep learning models in abstract visual reasoning.
ABNet: Mitigating Sample Imbalance in Anomaly Detection Within Dynamic Graphs
Yifan Hong (Huazhong Agricultural University), Di Wang (King Abdullah University of Science and Technology)
Anomaly DetectionMeta LearningGraph Neural NetworkAuto EncoderGraph
🎯 What it does: Proposed a framework named ABNet to address the problem of sample imbalance in anomaly node detection on dynamic graphs.
Abstraction Heuristics for Classical Planning Tasks with Conditional Effects
Martín Pozo (Universidad Carlos III de Madrid), Jendrik Seipp (Linkoping University)
OptimizationBenchmark
🎯 What it does: This paper proposes a general method that directly supports conditional effects in classical planning tasks, extending the implementations of pattern databases (PDB), merge-and-shrink (M&S), and Cartesian abstractions for admissible heuristics, and improving algorithms based on Counterexample-Guided Abstraction Refinement (CEGAR) to handle these tasks.
AccCtr: Accelerating Training-Free Conditional Control For Diffusion Models
Longquan Dai (Nanjing University of Science and Technology), Jinhui Tang (Nanjing University of Science and Technology)
GenerationOptimizationComputational EfficiencyDiffusion modelImage
🎯 What it does: Propose the AccCtr method, which significantly accelerates the sampling process by performing alternating maximization on training-free conditional diffusion models (CDM) (first unconditional optimization, then conditional optimization), and by retraining the conditional extraction network.
Accelerating Adversarial Training on Under-Utilized GPU
Zhuoxin Zhan (Simon Fraser University), Pulei Xiong (National Research Council Canada)
Computational EfficiencyAdversarial AttackImageTabular
🎯 What it does: Propose AttackRider, a method that leverages low GPU utilization to accelerate adversarial training;
Accelerating Diffusion-based Super-Resolution with Dynamic Time-Spatial Sampling
Rui Qin (Tsinghua University), Bin Wang (Tsinghua University)
Super ResolutionDiffusion modelImage
🎯 What it does: This paper proposes a TSS method for accelerating diffusion model-based image super-resolution without additional training, based on spatiotemporal adaptive sampling.
Accurate Sublayer Pruning for Large Language Models by Exploiting Latency and Tunability Information
Seungcheol Park (Seoul National University), U Kang (Seoul National University)
Computational EfficiencyTransformerLarge Language ModelText
🎯 What it does: This paper proposes a novel sublayer pruning method called SPRINT, specifically designed to enhance inference speed for large language models (LLMs) while maintaining accuracy.
Active Multimodal Distillation for Few-shot Action Recognition
Weijia Feng (Tianjin Normal University), Xiaobai Li (Zhejiang University)
RecognitionKnowledge DistillationMeta LearningConvolutional Neural NetworkOptical FlowVideoMultimodality
🎯 What it does: Proposed a multi-modal few-shot action recognition framework based on active inference, which actively identifies the dominant modality (RGB or optical flow) for each query sample, enhances the suboptimal modality's representation through bidirectional mutual distillation, and finally employs adaptive multi-modal reasoning for fusion during the testing phase.
ActiveHAI: Active Collection Based Human-AI Diagnosis with Limited Expert Predictions
Xuehan Zhao (Northwestern Polytechnical University), Bin Guo (Northwestern Polytechnical University)
ClassificationConvolutional Neural NetworkTransformerBiomedical DataElectronic Health Records
🎯 What it does: Proposed ActiveHAI, combining median window active sampling with an evaluation module to achieve efficient human-machine collaborative diagnosis when expert predictions are limited.
AdaMixT: Adaptive Weighted Mixture of Multi-Scale Expert Transformers for Time Series Forecasting
Huanyao Zhang (Peking University), Guoliang Li (Tsinghua University)
TransformerLarge Language ModelMixture of ExpertsTime Series
🎯 What it does: Designed a multi-scale expert mixture Transformer called AdaMixT for multivariate time series prediction;
Adaptive Deep Learning from Crowds
Hang Yang (Macau University of Science and Technology), Witold Pedrycz (University of Alberta)
ClassificationImage
🎯 What it does: Propose AdaCrowd, a probability model-based adaptive learning framework that can dynamically select the most informative instances during the label collection process, effectively estimating worker parameters and training the target model with only a small amount of annotations.
Adaptive Gradient Learning for Spiking Neural Networks by Exploiting Membrane Potential Dynamics
Jiaqiang Jiang (Zhejiang University of Technology), Rui Yan (Zhejiang University of Technology)
OptimizationComputational EfficiencySpiking Neural NetworkImage
🎯 What it does: This paper proposes an adaptive gradient learning framework (MPD-AGL), which dynamically adjusts the width of synaptic gradients to align with the distribution of membrane potential dynamics (MPD) across different time steps, thereby alleviating the problem of gradient vanishing or mismatch in SNN training.
Adaptive Graph Unlearning
Pengfei Ding (Macquarie University), Jiajie Zhu (Macquarie University)
Safty and PrivacyGraph Neural NetworkGraph
🎯 What it does: Propose AGU, a forgetting framework that adapts to different graph neural networks and various graph elements, achieving complete forgetting of nodes, edges, and features.
Adaptive Language-Aware Image Reflection Removal Network
Siyan Fang (Huazhong University of Science and Technology), Yuehuan Wang (Huazhong University of Science and Technology)
RestorationVision Language ModelMultimodalityBenchmark
🎯 What it does: Propose Adaptive Language-Aware Network (ALANet) for removing complex reflections in single images, even when the provided language descriptions contain errors;
Adaptive Wizard for Removing Cross-Tier Misconfigurations in Active Directory
Huy Q. Ngo (University of Adelaide), Hung X. Nguyen (University of Adelaide)
OptimizationReinforcement LearningGraph
🎯 What it does: Propose the Adaptive Path Removal (APR) problem and design precise, approximate, and scalable heuristic algorithms to guide IT administrators in progressively removing attack paths in Active Directory graphs using a wizard-based interface;
AdaptPFL: Unlocking Cross-Device Palmprint Recognition via Adaptive Personalized Federated Learning with Feature Decoupling
Zirui Zhang (Nanjing University of Aeronautics and Astronautics), Qi Zhu (Nanjing University of Aeronautics and Astronautics)
RecognitionDomain AdaptationFederated LearningSafty and PrivacyConvolutional Neural NetworkImage
🎯 What it does: Studied an adaptive personalized federated learning method based on feature decoupling to improve the accuracy and privacy protection of cross-device fingerprint recognition.
AdaR: An Adaptive Gradient Method with Cyclical Restarting of Moment Estimations
Yangchuan Wang (Beijing Wuzi University), Peng Shi (University of Science and Technology Beijing)
OptimizationImageText
🎯 What it does: Propose AdaR, which suppresses long-tail gradients by periodically restarting Adam's momentum estimation, thereby enhancing the optimizer's generalization and convergence speed.
ADC-GS: Anchor-Driven Deformable and Compressed Gaussian Splatting for Dynamic Scene Reconstruction
He Huang (Shanghai Jiao Tong University), Zhu Li (University of Missouri-Kansas City)
CompressionOptimizationComputational EfficiencyGaussian SplattingVideoPoint Cloud
🎯 What it does: Proposes an Anchor-Driven Deformable and Compressed Gaussian Splatting (ADC-GS) framework, achieving efficient compression and rendering of dynamic scenes through anchor-driven hierarchical deformation and multi-dimensional entropy models.
ADFormer: Aggregation Differential Transformer for Passenger Demand Forecasting
Haichen Wang (East China Normal University), Jilin Hu (East China Normal University)
TransformerTime Series
🎯 What it does: Designed and implemented a new Aggregated Differential Transformer (ADFormer) for urban passenger demand prediction, integrating spatial differential attention, spatial/temporal aggregation strategies, and hierarchical time matrices to improve prediction accuracy.
ADPFedGNN: Adaptive Decoupling Personalized Federated Graph Neural Network
Zeli Guan (Beijing University of Posts and Telecommunications), Xiaolong Meng (Beijing University of Posts and Telecommunications)
ClassificationFederated LearningGraph Neural NetworkReinforcement LearningGraph
🎯 What it does: Propose an adaptive decoupled personalized federated graph neural network, ADPFedGNN, for training localized node classification models in non-IID graph data environments;
Advancing Community Detection with Graph Convolutional Neural Networks: Bridging Topological and Attributive Cohesion
Anjali de Silva (Victoria University of Wellington), Xingquan Zuo (Beijing University of Posts and Telecommunications)
Graph Neural NetworkGraph
🎯 What it does: Propose a new graph convolutional network community detection framework called TAS-Com, which trains GCN by combining high-quality community structures generated by the Leiden algorithm with the topological connectivity of human-annotated communities, ultimately achieving better community partitioning in attributed graphs.
Advancing Embodied Agent Security: From Safety Benchmarks to Input Moderation
Ning Wang (Chongqing University), Tao Xiang (Chongqing University)
Safty and PrivacyTransformerLarge Language ModelPrompt EngineeringTextBenchmark
🎯 What it does: Proposes a complete input review framework for embodied agents, including the safety benchmark EAsafetyBench and a lightweight Prompt-decoupled review method called Pinpoint.
Advancing Generalization Across a Variety of Abstract Visual Reasoning Tasks
Mikołaj Małkiński (Warsaw University of Technology), Jacek Mańdziuk (Warsaw University of Technology)
ClassificationConvolutional Neural NetworkTransformerImageMeshBenchmark
🎯 What it does: Proposed and evaluated the Pathways of Normalized Group Convolution (PoNG) model, conducting experiments on the generalization performance of abstract visual reasoning (AVR) tasks (e.g., Raven Progressive Matrices, Visual Analogies) under both i.i.d. and out-of-distribution (o.o.d.) settings.
Advancing Stain Transfer for Multi-Biomarkers: A Human Annotation-Free Method Based on Auxiliary Task Supervision
Siyuan Xu (East China Normal University), Qingli Li (East China Normal University)
Image TranslationConvolutional Neural NetworkTransformerGenerative Adversarial NetworkBiomedical Data
🎯 What it does: Investigated an H&E to IHC staining transfer method that does not require manual annotation and is applicable to multiple biomarkers, capable of generating virtual IHC images while maintaining pathological and structural consistency.
Adversarial Attacks on Both Face Recognition and Face Anti-spoofing Models
Fengfan Zhou (Huazhong University of Science and Technology), Xuequan Lu (University of Western Australia)
RecognitionAdversarial AttackConvolutional Neural NetworkImage
🎯 What it does: A framework capable of simultaneously attacking both face recognition and anti-spoofing models in an integrated system is proposed.
Adversarial Propensity Weighting for Debiasing in Collaborative Filtering
Kuiyu Zhu (Xi'an Jiaotong University), Xin Wang (Xi'an Jiaotong University)
Recommendation SystemGraph Neural NetworkGenerative Adversarial NetworkTabular
🎯 What it does: Propose a collaborative filtering debiasing method APWCF that combines dynamic propensity modeling with adversarial learning to simultaneously eliminate popularity bias and selection bias;
Adversarial Training for Graph Convolutional Networks: Stability and Generalization Analysis
Chang Cao (Huazhong Agricultural University), Hong Chen (Huazhong Agricultural University)
Adversarial AttackGraph Neural NetworkGraphBenchmark
🎯 What it does: This paper derives the expected generalization error upper bound for graph convolutional networks (GCN) under adversarial training against node and structural attacks through uniform stability analysis;
AdvGrasp: Adversarial Attacks on Robotic Grasping from a Physical Perspective
Xiaofei Wang (University of Science and Technology of China), Keke Tang (Guangzhou University)
Adversarial AttackRobotic IntelligencePoint CloudMeshBenchmark
🎯 What it does: Proposes an adversarial attack framework called AdvGrasp from a physical perspective, distorting shapes to undermine lift capability and grasp stability in robotic grasping.
Aggregation Mechanism Based Graph Heterogeneous Networks Distillation
Xiaobin Hong (Nanjing University), Sanglu Lu (Nanjing University)
Knowledge DistillationGraph Neural NetworkTransformerGraph
🎯 What it does: Propose the AMEND framework, distilling complex graph neural networks (Graph Transformer) into lightweight multi-layer perceptrons (MLP), focusing on preserving and aligning aggregation mechanisms.
Airdrop Games
Sotiris Georganas (IOG), Paolo Penna (IOG)
OptimizationFinance Related
🎯 What it does: Model the airdrop mechanism as a game, analyze pure Nash and noise response equilibria, and provide the optimal airdrop scale.
AKBR: Learning Adaptive Kernel-based Representations for Graph Classification
Lu Bai (Beijing Normal University), Edwin Hancock (University of York)
ClassificationRepresentation LearningDrug DiscoveryGraph Neural NetworkGraph
🎯 What it does: Propose an end-to-end adaptive kernel representation (AKBR) framework that uses an attention mechanism to weight substructures in traditional R-convolution kernels (e.g., WLSK, SPGK), automatically learning the importance of substructures and constructing a learnable kernel matrix for graph classification tasks;
Aligning Contrastive Multiple Clusterings with User Interests
Shan Zhang (Shandong University), Guoxian Yu (Shandong University)
Convolutional Neural NetworkContrastive LearningImageTextBiomedical DataBenchmark
🎯 What it does: This paper proposes a multi-clustering framework CMClusts based on contrastive learning and user interest-guided data augmentation, capable of generating diverse clustering results aligned with different user interests.
All Roads Lead to Rome: Exploring Edge Distribution Shifts for Heterophilic Graph Learning
Yi Wang (Zhejiang Normal University), Xiaodi Huang (Charles Sturt University)
Representation LearningGraph Neural NetworkGraph
🎯 What it does: Propose a heterogeneous graph neural network framework called HOGNN, which achieves precise distinction between homogeneous and heterogeneous edges by treating edge attribute inference as an out-of-distribution (OOD) detection task, thereby enhancing node representation learning.
AlphaGAT: A Two-Stage Learning Approach for Adaptive Portfolio Selection
Shicheng Li (Wuhan University), Feng Wang (Wuhan University)
Graph Neural NetworkReinforcement LearningTime SeriesFinance Related
🎯 What it does: Propose the AlphaGAT two-phase learning framework for adaptive portfolio selection in non-static financial markets.
Always Clear Depth: Robust Monocular Depth Estimation Under Adverse Weather
Kui Jiang (Harbin Institute of Technology), Jingchun Zhou (Dalian Maritime University)
Depth EstimationAutonomous DrivingKnowledge DistillationDiffusion modelImageBenchmark
🎯 What it does: Designed a robust monocular depth estimation framework ACDepth aimed at solving depth estimation problems under adverse weather conditions.
An Approach to Quantify Plans Robustness in Real-world Applications
Francesco Percassi (University of Huddersfield), Mauro Vallati (University of Huddersfield)
OptimizationRobotic IntelligenceTabular
🎯 What it does: Proposes a statistical framework for evaluating plan robustness in uncertain environments, and introduces the concepts of execution-invariant tasks and B-robustness.
An Association-based Fusion Method for Speech Enhancement
Shijie Wang (Shanxi University), Xinyan Liang (Shanxi University)
RestorationConvolutional Neural NetworkGraph Neural NetworkAudio
🎯 What it does: Proposed a speech enhancement method called AFSE based on association fusion, which first uses a graph neural network to capture global frame associations, and then employs a dilated convolution UNet to extract local features, achieving high-quality speech enhancement.
An Efficient Core-Guided Solver for Weighted Partial MaxSAT
Shiwei Pan (Northeast Normal University), Shaowei Cai (Chinese Academy of Sciences)
OptimizationBenchmark
🎯 What it does: Proposes CASHWMaxSAT, a core-guided Weighted Partial MaxSAT solver that significantly improves solving speed while maintaining optimality.
An End-to-End Simple Clustering Hierarchical Pooling Operation for Graph Learning Based on Top-K Node Selection
Zhehan Zhao (Beijing Normal University), Edwin Hancock
ClassificationComputational EfficiencyGraph Neural NetworkGraph
🎯 What it does: Designed an end-to-end simple clustering hierarchical pooling method called SCHPool, which progressively compresses graphs through layer-by-layer compression using Top-K node selection combined with multi-perspective importance scoring and position attention, achieving O(E+N) computational complexity via sparse matrices.
An Inverse Optimization Approach to Contextual Inverse Optimization
Yasunari Hikima (Kyushu University), Naoyuki Kamiyama (Kyushu University)
Optimization
🎯 What it does: Proposes a general inverse optimization method for context inverse optimization (CIO) in scenarios where only historical optimal solutions are available instead of parameters, applicable to nonlinear problems such as integer programming;
An Out-Of-Distribution Membership Inference Attack Approach for Cross-Domain Graph Attacks
Jinyan Wang (Guangxi Normal University), Xingcheng Fu (Guangxi Normal University)
Adversarial AttackGraph Neural NetworkGraph
🎯 What it does: Designed a cross-domain membership inference attack framework called GOOD-MIA, leveraging OOD (Out-of-Distribution) techniques to perform membership inference on GNNs across graph data with different distributions.
Antibody Design and Optimization with Multi-scale Equivariant Graph Diffusion Models for Accurate Complex Antigen Binding
Jiameng Chen (Wuhan University), Wenbin Hu (Wuhan University Shenzhen Research Institute)
Drug DiscoveryProtein Structure PredictionGraph Neural NetworkDiffusion modelGraphBiomedical Data
🎯 What it does: Propose AbMEGD, an end-to-end multi-scale E(3) equivariant graph diffusion framework for the co-design and affinity optimization of antibody sequences and structures.
APIMig: A Project-Level Cross-Multi-Version API Migration Framework Based on Evolution Knowledge Graph
Li Kuang (Central South University), YingJie Xia (Hangzhou Dianzi University)
AI Code AssistantGraph Neural NetworkTransformerLarge Language ModelPrompt EngineeringTextGraph
🎯 What it does: Propose the APIMig framework to address cross-multi-version project-level API migration issues;
App2Exa: Accelerating Exact kNN Search via Dynamic Cache-Guided Approximation
Ke Li (University of Electronic Science and Technology of China), Shuo Shang (University of Electronic Science and Technology of China)
RetrievalComputational EfficiencyImageText
🎯 What it does: For scenarios involving low to medium-dimensional data with time-varying query distributions, App2Exa is proposed—a cache-guided framework that provides approximate results through dynamic cache graph indexing and then performs exact kNN search using VP-Tree, significantly improving the query efficiency of exact kNN.
Approximate Lifted Model Construction
Malte Luttermann (German Research Center for Artificial Intelligence), Mattis Hartwig (German Research Center for Artificial Intelligence)
Computational EfficiencyRepresentation LearningGraphBiomedical DataElectronic Health Records
🎯 What it does: Proposed the ε-ACP algorithm to construct approximate extractable representations in probabilistic relational models by allowing slight deviations in potential functions, thereby enabling efficient extractable inference.
Approximate Verification of Strategic Abilities under Imperfect Information Using Local Models
Damian Kurpiewski (Polish Academy of Sciences), Yan Kim (University of Luxembourg)
🎯 What it does: This paper proposes a scheme that utilizes proxy local models for approximation verification to check strategy capabilities under incomplete information conditions.
Approximated Behavioral Metric-based State Projection for Federated Reinforcement Learning
Zengxia Guo (China University of Petroleum-Beijing), Zhongqi Lu (China University of Petroleum-Beijing)
Federated LearningSafty and PrivacyReinforcement LearningRetrieval-Augmented Generation
🎯 What it does: Propose the FedRAG framework in federated reinforcement learning, utilizing parameter sharing of state projection functions based on approximate behavioral metrics to enhance policy learning across different environments;
Approximately EFX and fPO Allocations for Bivalued Chores
Zehan Lin (University of Macau), Shengwei Zhou (University of Macau)
Optimization
🎯 What it does: This paper proposes a polynomial-time algorithm that can compute a (2-1/k)-EFX allocation satisfying fPO for bivalued chores, and further achieves fully EFX and fPO allocations when k=2.
Approximation Fixpoint Theory as a Unifying Framework for Fuzzy Logic Programming Semantics
Pascal Kettmann (TU Dresden), Hannes Strass (TU Dresden)
Review/Survey Paper
🎯 What it does: Reconstructed the two major semantics of fuzzy logic programs (stable model semantics and approximate infinite depth semantics) into the Approximating Fixed Point Theory (AFT) framework, unified the two semantics, and derived their relationships.
Are Large Language Models Fluent in Declarative Process Mining?
Valeria Fionda (University of Calabria), Francesco Ricca (University of Calabria)
TransformerLarge Language ModelTextBenchmark
🎯 What it does: Developed and evaluated a framework for quantitative assessment of bidirectional translation between natural language and Declare declarative process models.
Argument-based Multi-Issue Negotiation
Thalya Fossey (Université Paris Cités), Pavlos Moraitis (Université Paris Cités)
🎯 What it does: Propose a hybrid framework that combines utility-driven multi-issue negotiation with argumentation-based constraints to achieve both rapid convergence and ensure proposals align with the beliefs of all parties.
AriGraph: Learning Knowledge Graph World Models with Episodic Memory for LLM Agents
Petr Anokhin (AIRI), Evgeny Burnaev (Skoltech)
Graph Neural NetworkTransformerLarge Language ModelAgentic AIWorld ModelTextGraphRetrieval-Augmented Generation
🎯 What it does: Propose AriGraph memory graph, integrating semantic knowledge graphs with episodic memory to build the LLM agent Ariadne, supporting long-term memory and decision-making in text games and multi-hop question answering.
ARMR: Adaptively Responsive Network for Medication Recommendation
Feiyue Wu (Southeast University), Shenqi Jing (Nanjing Medical University)
Recommendation SystemTransformerTabularTime SeriesBiomedical DataElectronic Health Records
🎯 What it does: Propose the ARMR adaptively responsive network for dynamic modeling of historical and new drug recommendations, enhancing personalized medication suggestions.
ARPDL: Adaptive Relational Prior Distribution Loss as an Adapter for Document-Level Relation Extraction
Huangming Xu (Northeastern University), Xin Li (Northeastern University)
Representation LearningTransformerLarge Language ModelText
🎯 What it does: Propose a new multi-label loss function called ARPDL for document-level relation extraction tasks, and apply it as a generic adapter to existing threshold loss functions;
ASCENT-ViT: Attention-based Scale-aware Concept Learning Framework for Enhanced Alignment in Vision Transformers
Sanchit Sinha (University of Virginia), Aidong Zhang (University of Virginia)
ClassificationExplainability and InterpretabilityRepresentation LearningTransformerImageBiomedical Data
🎯 What it does: Propose the ASCENT-ViT framework, integrating multi-scale encoding, deformable multi-scale fusion, and concept alignment modules into Vision Transformer to achieve interpretable concept learning and prediction.
Assessing the Exposure to Public Knowledge in Policy-Protected Description Logic Ontologies
Gianluca Cima (Sapienza University of Rome), Domenico Fabio Savo (University of Bergamo)
Safty and Privacy
🎯 What it does: This paper proposes a general framework to assess the risk of confidential information (protected by policy P) being leaked under public knowledge (K_pub), and defines two decision problems: existence of leakage (LE) and existence of a secure view (L-VE).