AAAI 2026 Papers — Page 14
AAAI Conference on Artificial Intelligence · 4149 papers
Facility Location for Congesting Commuters and Generalizing the Cost-Distance Problem
Thanasis Lianeas, Aikaterini Nikolidaki (University of West Attica)
Optimization
🎯 What it does: Proposed and studied a facility location problem where congestion affects connection costs, defining two new models, FLCC and FLSC, and providing approximation algorithms.
Fact2Fiction: Targeted Poisoning Attack to Agentic Fact-checking System
Haorui He (Hong Kong Baptist University), Francis C. M. Lau (University of Hong Kong)
Adversarial AttackTransformerLarge Language ModelAgentic AITextRetrieval-Augmented Generation
🎯 What it does: Proposed a poisoning attack framework called Fact2Fiction targeting agent-based fact-checking systems, generating customized malicious evidence for sub-claims by leveraging justifications produced by the system;
FACTGUARD: Event-Centric and Commonsense-Guided Fake News Detection
Jing He (Yunnan University), Renyang Liu (National University of Singapore)
ClassificationExplainability and InterpretabilityComputational EfficiencyKnowledge DistillationTransformerLarge Language ModelText
🎯 What it does: Propose the FACTGUARD framework, which leverages LLMs to extract core event content and combines common-sense reasoning for fake news detection.
Factorization-in-Loop:Proximal Fill-in Minimization for Sparse Matrix Reordering
Ziwei Li (Institute of Software Chinese Academy of Sciences), Wenjia Wu (Institute of Software Chinese Academy of Sciences)
OptimizationGraph Neural NetworkGraphBenchmark
🎯 What it does: Propose a Proximal Fill-in Minimization (PFM) framework based on graph neural networks, which jointly learns to minimize fill-in through a differentiable reordering layer and factorization-enhanced loss before LU decomposition of sparse matrices.
Failure Localization in Multi-Agent Code Generation via Knowledge-Guided and Transferable Reasoning
Mingyang Geng (National University of Defense Technology), Haotian Wang (Northeastern University)
AI Code AssistantTransformerContrastive LearningTextBenchmark
🎯 What it does: Propose FLKR—an unsupervised multi-agent code generation failure localization framework, and construct COFL, a fine-grained expert-annotated benchmark.
Failures to Surface Harmful Contents in Video Large Language Models
Yuxin Cao, Jin Song Dong (Csiro's Data61)
Safty and PrivacyAdversarial AttackTransformerLarge Language ModelVision Language ModelVideo
🎯 What it does: Analyze and reveal the blind spots of video large language models (LLMs) in detecting harmful content, and validate this vulnerability through constructing three no-query black-box attacks.
Fair Algorithms with Probing for Multi-Agent Multi-Armed Bandits
Tianyi Xu (Tulane University), Zizhan Zheng (Tulane University)
OptimizationReinforcement LearningTabularTime Series
🎯 What it does: This paper proposes a multi-agent multi-armed bandit (MA-MAB) framework and introduces a probing mechanism, which first samples several arms in each round and then performs fair allocation based on Nash Social Welfare (NSW).
Fair Allocation of Indivisible Goods with Variable Groups
Paul Gölz (Cornell University), Warut Suksompong (National University of Singapore)
Optimization
🎯 What it does: Under the variable group model (where participants can be freely divided into multiple groups and discrete items can be allocated), it is proven that regardless of the number of groups or group sizes, an EF1 (envy-free up to one item) allocation exists; an algorithm based on an extended envy cycle elimination method is proposed to achieve this result; further, it is proven that if items are arranged on a path and each group receives a connected bundle, a connected EF1 allocation can also be obtained; under random additive utilities, the asymptotic existence of EF is studied, with thresholds provided for both divisible and indivisible cases (Θ(log n) and Ω(√n), respectively).
Fair and Efficient Balanced Allocation for Indivisible Goods
Yasushi Kawase (University of Tokyo), Ryoga Mahara (University of Tokyo)
Optimization
🎯 What it does: Under the balance constraint (each agent receives the same number of indivisible goods), the problem of achieving fair (EF1) and efficient (fPO) allocation is studied, with proofs of existence and polynomial-time algorithms provided for two scenarios (individual bilateral preferences and up to two types of agents).
Fair Bayesian Data Selection via Generalized Discrepancy Measures
Yixuan Zhang (Southeast University), Quyu Kong (Alibaba Cloud)
Data-Centric LearningMeta LearningImage
🎯 What it does: Propose a fair learning framework based on Bayesian data selection, achieving data-level fairness by aligning grouped posteriors with central distributions.
Fair Diffusion Auctions
Zixin Gu (Shanghaitech University), Dengji Zhao (Shanghaitech University)
🎯 What it does: Designed a fair diffusion auction mechanism in social networks, proposing Permutation Diffusion Auction (PDA) and its combination extension CPDA.
Fair Division Among Couples and Small Groups
Paul Gölz (Cornell University), Hannane Yaghoubizade (Cornell University)
Optimization
🎯 What it does: The paper studies the fair division problem of allocating indivisible items among couples and small groups, proving that EF1 is achievable between two couples, but not necessarily for three or more couples.
Fair Domain Generalization: An Information-Theoretic View
Tangzheng Lian (King's College London), Oya Celiktutan (Queen Mary University of London)
Domain AdaptationImageText
🎯 What it does: Propose the FairDG framework, providing upper bounds on the mutual information between expected risk and fairness violation, and achieving a trade-off between accuracy and fairness through Pareto optimization.
Fair Facial Attribute Recognition via Group-Decoupled Vision Transformer with Mask-Guided Correlation Suppression
Huichang Huang (Xiamen University of Technology), Da-Han Wang (Xiamen University of Technology)
RecognitionTransformerImage
🎯 What it does: This paper proposes a group-decoupled Vision Transformer (GD-ViT) and a two-stage Mask-Guided Correlation Suppression learning strategy for fair facial attribute recognition;
Fair Incentives for Early Arrival in 0-1 Cooperative Games
Yaoxin Ge (ShanghaiTech University), Dengji Zhao (ShanghaiTech University)
Optimization
🎯 What it does: Propose a new fair allocation mechanism for online cooperative games, defining novel fairness metrics (Shapley distance and equilibrium welfare), achieving early-arrival incentive, Shapley fairness, online individual rationality, and symmetric player monotonicity in 0-1 monotonic games.
Fair Model-based Clustering
Jinwon Park (Seoul National University), Yongdai Kim (Seoul National University)
OptimizationComputational EfficiencyTabular
🎯 What it does: Proposed a fair clustering algorithm FMC based on finite mixture models, achieving fairness constraints through parameterized soft assignment, with the number of parameters independent of sample size.
Fair Societies: Algorithms for House Allocations
Hadi Hosseini (Pennsylvania State University), Aditi Sethia (Indian Institute of Science)
Optimization
🎯 What it does: Proposes two classes of algorithms to minimize the number of envious agents in housing allocation problems: one is a metric-agnostic FPT improvement framework for a given initial allocation, and the other is polynomial/linear time algorithms under single-peaked/single-concave preference domains; and experiments verify the impact of reallocation on fairness and welfare.
FairGC: Fostering Individual and Group Fairness for Deep Graph Clustering
Haodong Zhang (Northeastern University), Wei Ju (Chinese Academy of Sciences)
Graph Neural NetworkContrastive LearningGraph
🎯 What it does: Proposes a fair deep graph clustering framework, FairGC, which jointly achieves individual fairness and group fairness to enhance clustering quality;
FairGSE: Fairness-Aware Graph Neural Network Without High False Positive Rates
Zhenqiang Ye (Jinan University), Xuemin Wang (Guilin University of Electronic Technology)
Graph Neural NetworkContrastive LearningGraphTabularFinance Related
🎯 What it does: This paper addresses the high false positive rate (FPR) problem in fair graph neural networks by proposing a fairness-enhancing framework called FairGSE based on two-dimensional structural entropy (2D-SE).
Fairness and Stability for Shared Resource Allocation Problems
Jiazhu Fang (Ocean University of China), Wenjing Liu (Ocean University of China)
Optimization
🎯 What it does: Studies the shared resource allocation problem, defining multiple concepts of fairness and stability (MMS, EF, NS, EEF, ENS, SS), and analyzes their existence and computational complexity under two categories of value monotonicity: non-increasing and non-decreasing.
Fairness Aware Reinforcement Learning via Proximal Policy Optimization
Gabriele La Malfa (King's College London), Elizabeth Black (King's College London)
Reinforcement Learning
🎯 What it does: A fair reinforcement learning algorithm called Fair-PPO based on Proximal Policy Optimization (PPO) is proposed for multi-agent systems. It introduces two penalty terms (retrospective and prospective) based on group fairness metrics (e.g., demographic parity), encouraging agents to pursue reward maximization while satisfying fairness.
Fairness in Repeated Matching: A Maximin Perspective
Eugene Lim (National University of Singapore), Nicholas Teh (National University of Singapore)
Optimization
🎯 What it does: Studies the decision problem of achieving fairness (maximin) objectives in multi-round matching, exploring its feasibility and complexity;
Fairness in the Multi-Secretary Problem
Georgios Papasotiropoulos (University of Warsaw), Zein Pishbin (University of Warsaw)
OptimizationTabular
🎯 What it does: This paper studies the introduction of fairness objectives in online decision-making for the multi-secretary problem, proposing and analyzing online mechanisms based on social choice rules.
Fairness Perceptions of Large Language Models
Benjamin Cookson (University of Toronto), Nisarg Shah (University of Toronto)
OptimizationTransformerLarge Language ModelPrompt Engineering
🎯 What it does: Assess the perception of fairness and decision-making behavior of large language models in fair allocation tasks.
Fairness-Aware Design for Contextual Experiments: Guaranteeing Reliability and Equity in Heterogeneous Subgroups
Guangyan Gan (Nanyang Technological Unviersity), Peng Jiang (Kuaishou Technology)
OptimizationTabularBiomedical Data
🎯 What it does: Propose a fair-aware contextual Thompson sampling design (F-CTSD) algorithm for experimental design in heterogeneous subgroups, ensuring subgroup fairness and statistical reliability.
Faithful in Steps: Improving Generalization and Citation in RAG via Query Decomposition
Yue Liu (Hong Kong University of Science and Technology), Xiaofang Zhou (Hong Kong University of Science and Technology)
RetrievalTransformerSupervised Fine-TuningVision Language ModelMultimodalityRetrieval-Augmented GenerationChain-of-Thought
🎯 What it does: Propose a RAG framework called QDRAG based on query decomposition, which can achieve accurate citations and trustworthy answers in multi-hop, multi-modal question answering.
False Positives Matter: Multidimensional Localization Evaluation and Training-Free Explainable Adversarial Patch Defense
Lihua Jing (Institute of Information Engineering, Chinese Academy of Sciences), Zixuan Zhu (Institute of Information Engineering, Chinese Academy of Sciences)
Explainability and InterpretabilityAdversarial AttackTransformerLarge Language ModelDiffusion modelImageVideoMultimodalityChain-of-Thought
🎯 What it does: This paper proposes a multi-dimensional evaluation framework and designs a semantics-based, training-free, interpretable adversarial patch defense method called SATED;
FAM: Fine-Grained Alignment Matters in Multimodal Embedding Learning with Large Vision-Language Models
Tianhang Xiang (South China University of Technology), Mingkui Tan (South China University of Technology)
Object DetectionSegmentationRetrievalRepresentation LearningTransformerVision Language ModelContrastive LearningImageTextMultimodality
🎯 What it does: This paper proposes the FAM (Fine-grained Alignment Matters) framework, transforming large-scale vision-language models (LVLM) into high-quality multimodal embedding models. The framework involves two-stage training: ① Alignment phase: Multi-grained Alignment Contrastive (MAC) loss is used for coarse-to-fine feature alignment on image-text pairs; ② Adaptation phase: Vision Embedding Inversion (VEIN) performs masked reconstruction on visual features, encouraging embeddings to retain fine-grained visual information, followed by contrastive learning on downstream multimodal retrieval/localization tasks.
FAMDR: Feature-Aligned Multimodal Denoising for Reliable Diagnostic Reconciliation in Medical Imaging
Xun Liang (Renmin University of China), Hongxun Jiang (Renmin University of China)
Anomaly DetectionTransformerContrastive LearningImageTextMultimodalityBiomedical DataElectronic Health RecordsRetrieval-Augmented Generation
🎯 What it does: Developed the FAMDR framework for reliable diagnostic alignment and noise elimination between medical imaging and electronic medical records.
FaNe: Towards Fine-Grained Cross-Modal Contrast with False-Negative Reduction and Text-Conditioned Sparse Attention
Peng Zhang, Heng Kong (Shenzhen University)
ClassificationObject DetectionSegmentationTransformerVision Language ModelContrastive LearningMultimodalityBiomedical DataBenchmark
🎯 What it does: This study proposes the FaNe framework, addressing issues of pseudo-negative samples, coarse-grained alignment, and intra-modal discriminability in medical image vision-language pre-training through semantically enhanced positive sample mining, text-conditioned sparse attention pooling, and hard negative sample adaptive contrastive loss.
FANoise: Singular Value-Adaptive Noise Modulation for Robust Multimodal Representation Learning
Jiaoyang Li (JD, Retail), Qixia Jiang (JD, Retail)
Representation LearningContrastive LearningMultimodality
🎯 What it does: Proposed a feature-adaptive noise injection method based on singular value decomposition (FANoise) to improve multimodal representation learning.
FantasyHSI: Video-Generation-Centric 4D Human Synthesis in Any Scene Through a Graph-Based Multi-Agent Framework
Lingzhou Mu (Tsinghua University), Kai Zhang (AMAP, Alibaba Group)
Data SynthesisGraph Neural NetworkReinforcement LearningAgentic AIVision Language ModelDiffusion modelVideo
🎯 What it does: Designed the FantasyHSI framework to achieve 4D human-computer interaction based on video generation, using a multi-agent system to generate long temporal sequences of animations with human behavior and physical consistency in arbitrary 3D scenes.
FantasyStyle: Controllable Stylized Distillation for 3D Gaussian Splatting
Yitong Yang (Shanghai University of Finance and Economics), Shuting He (Shanghai University of Finance and Economics)
GenerationKnowledge DistillationDiffusion modelScore-based ModelGaussian Splatting
🎯 What it does: Proposes the FantasyStyle framework, which utilizes 3D Gaussian Splatting for style transfer in 3D scenes, and addresses multi-view inconsistency and content leakage issues through frequency filtering and negative guidance.
FantasyTalking2: Timestep-Layer Adaptive Preference Optimization for Audio-Driven Portrait Animation
Mengchao Wang (Alibaba Group), Mu Xu (Alibaba Group)
GenerationReinforcement Learning from Human FeedbackTransformerLarge Language ModelReinforcement LearningMixture of ExpertsDiffusion modelVideoMultimodalityAudio
🎯 What it does: Proposed a Timestep-Layer Adaptive Preference Optimization (TLPO) framework combined with a multimodal reward model called Talking-Critic, enhancing the performance of audio-driven portrait animation in terms of motion naturalness, lip synchronization, and visual quality.
FARM: Frame-Accelerated Augmentation and Residual Mixture-of-Experts for Physics-Based High-Dynamic Humanoid Control
Jing Tan (Hong Kong University of Science and Technology), Renjing Xu (Guangdong University of Technology)
Robotic IntelligenceTransformerReinforcement LearningMixture of ExpertsMultimodalitySequentialBenchmarkPhysics Related
🎯 What it does: Propose the FARM framework, integrating frame acceleration augmentation with residual Mixture-of-Experts (MoE), for unified control of high-dynamic humanoid motions.
Fashion Microscope: Pixel-Level Attribute Perception via Optimal Transport and Neural Semantic Aggregation
Shuili Zhang (Chinese Academy of Sciences), Tingwen Liu (Chinese Academy of Sciences)
RetrievalVision Language ModelContrastive LearningImage
🎯 What it does: Pro Fashion achieves a pixel-level attribute-aware fashion retrieval framework through hierarchical region→patch→superpixel segmentation, combined with optimal transport and neural semantic aggregation.
FashionMAC: Deformation-Free Fashion Image Generation with Fine-Grained Model Appearance Customization
Rong Zhang (Zhejiang Gongshang University), Xun Wang (Zhejiang Gongshang University)
Image TranslationGenerationTransformerDiffusion modelImage
🎯 What it does: Propose the FashionMAC framework, which directly generates extrapolated images from worn images using a deformation-free clothing center generation method to produce high-quality fashion display images;
Fast Conformal Prediction Using Conditional Interquantile Intervals
Naixin Guo (City University of Hong Kong), Zhixin Zhou (Alpha Benito Research)
Computational EfficiencyTabularBiomedical Data
🎯 What it does: Propose Conformal Interquantile Regression (CIR) and its improved version CIR+, which directly construct prediction intervals satisfying coverage probability through black-box quantile regression, avoiding histogram binning in traditional methods while balancing coverage and interval width.
Fast Guaranteed Robust Local-Smooth Principal Component Separation
Mingdi Hu (Xi'an University of Posts and Telecommunications), Jiangjun Peng (Northwestern Polytechnical University)
RestorationOptimizationImageVideo
🎯 What it does: Propose a new representative coefficient related total variation (RCCTV) regularizer that combines low-rank and local smoothness for robust principal component analysis;
Fast Multi-view Consistent 3D Editing with Video Priors
Liyi Chen (Hong Kong Polytechnic University), Lei Zhang (Hong Kong Polytechnic University)
GenerationComputational EfficiencyDiffusion modelGaussian SplattingVideoPoint Cloud
🎯 What it does: Leverage the time consistency prior of pre-trained video generation models to achieve multi-view consistent 3D editing with a single forward pass;
FastAnimate: Towards Learnable Template Construction and Pose Deformation for Fast 3D Human Avatar Animation
Jian Shu (Hong Kong University of Science and Technology (Guangzhou)), Hao Wang (Hong Kong University of Science and Technology (Guangzhou))
GenerationComputational EfficiencyConvolutional Neural NetworkGaussian SplattingMesh
🎯 What it does: Proposes a unified fast human avatar animation framework called FastAnimate based on 3D Gaussian splats, including fast template construction and a learnable pose deformation module.
FastDriveVLA: Efficient End-to-End Driving via Plug-and-Play Reconstruction-based Token Pruning
Jiajun Cao (Peking University), Shanghang Zhang (Peking University)
Autonomous DrivingTransformerVision-Language-Action ModelAuto EncoderGenerative Adversarial NetworkImage
🎯 What it does: Propose the FastDriveVLA framework, which uses ReconPruner to perform foreground information trimming on visual tokens of Vision-Language-Action models through MAE-style pixel reconstruction, significantly reducing the number of tokens.
Faster Certified Symmetry Breaking Using Orders with Auxiliary Variables
Markus Anders (RPTU Kaiserslautern-Landau), Yong Kiam Tan (Nanyang Technological University)
Computational EfficiencyBenchmark
🎯 What it does: This paper proposes using auxiliary variables to encode orderings in proof systems with verifiable symmetry breaking to achieve more efficient proof recording and checking.
Faster Game Solving via Asymmetry of Step Sizes
Linjian Meng (Nanjing University), Yang Gao (Nanjing University)
OptimizationReinforcement LearningBenchmark
🎯 What it does: In two-player zero-sum imperfect information games, Asymmetric PCFR+ (APCFR+) and its simplified version Simple APCFR+ (SAPCFR+) are proposed, enhancing the robustness of PCFR+ by introducing adaptive or fixed step size asymmetry between implicit and explicit cumulative counterfactual regret updates.
Faster Game Solving via Hyperparameter Schedules
Naifeng Zhang (Carnegie Mellon University), Tuomas Sandholm (Carnegie Mellon University)
OptimizationReinforcement LearningBenchmark
🎯 What it does: The study uses dynamic hyperparameter scheduling without training to improve the CFR algorithm in imperfect information games, significantly accelerating convergence.
Faster Symmetry Breaking Constraints for Abstract Structures
Özgür Akgün (University of St Andrews), Christopher Jefferson (University of Dundee)
Optimization
🎯 What it does: Proposed a symmetry breaking method for abstract structures, using representation-dependent ordering and delaying symmetry application to reduce constraint complexity.
FastFLUX: Pruning FLUX with Block-wise Replacement and Sandwich Training
Fuhan Cai (Shanghai Jiao Tong University), Xiangzhong Fang (Chinese University of Hong Kong)
GenerationComputational EfficiencySupervised Fine-TuningDiffusion modelImage
🎯 What it does: Perform architecture-level pruning on the FLUX text-to-image generation model, proposing the FastFLUX framework. It replaces the residual branch of ResBlock with linear layers using Block-wise Replacement with Linear Layers (BRLL), and adopts Sandwich Training (ST) for local fine-tuning to maintain image quality while significantly improving inference speed.
Fault Diagnosis of Irregular Sequences by Adjoint Learning in Continuous-Time Model Space
Xiren Zhou (University of Science and Technology of China), Huanhuan Chen (University of Science and Technology of China)
Anomaly DetectionRecurrent Neural NetworkTime SeriesOrdinary Differential Equation
🎯 What it does: Propose a method for modeling and fault diagnosis of irregularly sampled sequences in continuous-time model space, using CT-Res (ODE-embedded reservoir) to map sequences into a compact readout model, followed by efficient training through adjoint ESN with shared parameters and classification within the model space;
FCMO: A Flow-Curv Mamba Operator for Large-Scale 3D Vehicle Aerodynamics
Yuchen Xie (Nanjing Forestry University), Hengyi Ren (Nanjing Forestry University)
OptimizationComputational EfficiencyMeshPhysics Related
🎯 What it does: Propose a FlowCurv Mamba operator (FCMO) tailored for large-scale 3D automotive aerodynamics to efficiently predict surface pressure, wall shear stress, and drag coefficient.
FD-MAGRPO: Functionality-Driven Multi-Agent Group Relative Policy Optimization for Analog-LDO Sizing
Haoning Jiang (Southern University of Science and Technology), Junmin Jiang (Southern University of Science and Technology)
OptimizationReinforcement LearningTabular
🎯 What it does: Proposed a function-driven multi-agent group relative policy optimization (FD-MAGRPO) algorithm for automated size optimization of low-dropout regulator (LDO) circuits in simulation.
FDC-Ground: Improving GRPO for GUI Grounding via Exponential Rewards and Fact-Aligned Pruning
Xiangjian Zeng (Xiamen University), Liang Zhang (Xiaohongshu Inc)
TransformerReinforcement LearningImage
🎯 What it does: Propose the FDC-Ground framework, using reinforcement learning to address GUI localization tasks, with improved reward design and pruning strategies;
FDP: A Frequency-Decomposition Preprocessing Pipeline for Unsupervised Anomaly Detection in Brain MRI
Hao Li (Xiamen University), Liansheng Wang (Xiamen University)
Anomaly DetectionDiffusion modelAuto EncoderBiomedical DataMagnetic Resonance Imaging
🎯 What it does: Propose an unsupervised anomaly detection framework based on frequency-domain decomposition preprocessing (FDP), suppressing lesions through low-frequency reconstruction and preserving structural information via high-frequency components to achieve brain MRI anomaly detection;
Feature Attribution for Human Sensing with Radio Signals
Shuokang Huang (Imperial College London), Julie McCann (Imperial College London)
RecognitionExplainability and InterpretabilityData-Centric LearningTime Series
🎯 What it does: Propose and implement MatryMask, a feature attribution method for WiFi CSI human perception tasks that can explain the model's attention to human-related features in wireless signals.
Feature Integration Spaces: Joint Training Reveals Dual Encoding in Neural Network Representations
Omar Claflin (Independent Researcher)
Representation LearningAuto EncoderText
🎯 What it does: Investigate and propose a dual encoding space, employing jointly trained sparse autoencoders and neural factor machines (NFM) to simultaneously learn feature identity and feature integration effects, and validate their improvements in reconstruction and behavioral performance through intervention experiments.
Feature-Aware One-Shot Federated Learning via Hierarchical Token Sequences
Shudong Liu (Peking University), Guibo Luo (Peking University)
Federated LearningKnowledge DistillationTransformerImageBiomedical Data
🎯 What it does: Proposed a one-round communication federated learning framework called FALCON, which generates multi-scale token sequences through hierarchical scale encoding and constructs a global model using a multi-scale autoregressive Transformer generator and knowledge distillation.
Feature-Centric Unsupervised Node Representation Learning Without Homophily Assumption
Sunwoo Kim (KAIST), Kijung Shin (KAIST)
Representation LearningGraph Neural NetworkGraph
🎯 What it does: Propose an unsupervised node representation learning framework named FUEL without the equal assumption hypothesis, which can adaptively determine the degree of graph convolution usage to obtain high-quality node embeddings on graphs with varying levels of homogeneity.
FedAdamW: A Communication-Efficient Optimizer with Convergence and Generalization Guarantees for Federated Large Models
Junkang Liu, Zhouchen Lin (Tianjin University)
OptimizationFederated LearningConvolutional Neural NetworkTransformerSupervised Fine-TuningImageText
🎯 What it does: Proposed FedAdamW, an adaptive optimizer for large model training in federated learning, addressing issues of high variance in second-moment estimation, client drift caused by local overfitting, and slow convergence during reinitialization in each round.
FedALT: Federated Fine-Tuning Through Adaptive Local Training with Rest-of-World LoRA
Jieming Bian (University of Florida), Jie Xu (Middle Tennessee State University)
Federated LearningComputational EfficiencyTransformerLarge Language ModelMixture of ExpertsTextBenchmark
🎯 What it does: Propose FedALT, a personalized federated fine-tuning framework based on LoRA, which allows each client to continue training its own LoRA locally and introduces a frozen Rest-of-World (RoW) LoRA shared globally for knowledge aggregation.
FedARKS: Federated Aggregation via Robust and Discriminative Knowledge Selection and Integration for Person Re-identification
Xin Xu (Wuhan University of Science and Technology), Wei Liu (Wuhan University of Science and Technology)
RecognitionPose EstimationDomain AdaptationFederated LearningConvolutional Neural NetworkTransformerContrastive LearningImage
🎯 What it does: Propose the FedARKS framework under federated learning to address cross-domain generalization issues in person re-identification;
FedAU2: Attribute Unlearning for User-Level Federated Recommender Systems with Adaptive and Robust Adversarial Training
Yuyuan Li (Hangzhou Dianzi University), Chaochao Chen (Ant Group)
Recommendation SystemFederated LearningAuto EncoderTabular
🎯 What it does: Propose FedAU2 for attribute unlearning in user-level federated recommendation systems, combining adaptive adversarial training and dual stochastic variational autoencoders (DSVAE) to eliminate attribute information in user embeddings and prevent gradient leakage.
FedBRICK: Structural Bias Aware Heterogeneous Foundation Model Federated Tuning
Yuhang Zhang (Chinese University of Hong Kong Shenzhen), Fangxin Wang (Chinese University of Hong Kong Shenzhen)
Domain AdaptationFederated LearningKnowledge DistillationTransformerSupervised Fine-TuningImage
🎯 What it does: This study addresses the structural bias problem caused by partial training in deep models during model-heterogeneous federated optimization, proposing the FedBRICK framework to reduce client drift and enhance global performance.
FedCD: Towards Consolidated Distillation for Heterogeneous Federated Learning
Yichen Li (Huazhong University of Science and Technology), Imran Razzak (Mohamed bin Zayed University of Artificial Intelligence)
Federated LearningKnowledge DistillationConvolutional Neural NetworkImage
🎯 What it does: Proposed the FedCD method, which integrates feature distillation and logit distillation, employs cross-layer attention for adaptive feature aggregation, models feature distribution using a Gaussian Mixture Model (GMM), and achieves robust knowledge distillation in federated learning under heterogeneous environments.
FedCure: Mitigating Participation Bias in Semi-Asynchronous Federated Learning with Non-IID Data
Yue Chen (Wuhan University of Science and Technology), Guanghui Wen (Wuhan University of Science and Technology)
OptimizationFederated LearningImage
🎯 What it does: Propose the FedCure framework to address the participation bias caused by non-IID data in semi-asynchronous federated learning.
FedDNA: DNA Sequence Reconstruction via Deep Evidential Learning and Personalized Federated Aggregation
Haiyan Lin (Tianjin University), Yuping Duan (Tianjin University)
RestorationFederated LearningRecurrent Neural NetworkBiomedical Data
🎯 What it does: Propose the FedDNA framework for DNA sequence reconstruction in distributed DNA storage environments through federated learning.
Federated CLIP for Resource-Efficient Heterogeneous Medical Image Classification
Yihang Wu (Guilin University of Electronic Technology), Ahmad Chaddad (Ecole de Technologie Superieure)
ClassificationCompressionFederated LearningKnowledge DistillationTransformerVision Language ModelBiomedical Data
🎯 What it does: Propose a federated learning framework FedMedCLIP based on CLIP for low-resource, heterogeneous medical image classification.
Federated Context-Aware Personalized Recommendation
Zhihao Wang (Wuhan University), Bing Li (Wuhan University)
Recommendation SystemFederated LearningTransformerContrastive LearningSequential
🎯 What it does: Proposed a federated learning framework FedCAR, which constructs context representations using recent interaction sequences at clients and achieves personalized recommendation through contrastive learning, addressing issues of static user embeddings, poor interpretability, and data sparsity in traditional methods.
Federated Graph-level Clustering Network with Attribute Inference
Renda Han (Hainan University), Jieren Cheng (Hainan University)
Federated LearningGraph Neural NetworkGraph
🎯 What it does: Propose FedAI, integrating graph-level clustering and attribute inference to address knowledge drift caused by missing node attributes on clients.
Federated Incomplete Multi-View Clustering with Tensorized Low-Rank Constraint
Wei Feng (Northwest A&F University), Bin Liu (Northwest A&F University)
OptimizationFederated LearningRepresentation LearningImageTextMultimodality
🎯 What it does: Proposed and implemented a federated incomplete multi-view clustering framework called FIMVC-TLRC, which improves efficiency using anchor graphs and achieves globally consistent clustering representations through tensor low-rank constraints;
Federated Linear Dueling Bandits
Xuhan Huang, Zhongxiang Dai (Chinese University of Hong Kong)
Recommendation SystemFederated LearningTabular
🎯 What it does: Proposes a Federated Linear Dueling Bandits (FLDB) algorithm under the federated learning framework to address scenarios where multiple agents collaboratively learn without sharing raw data.
Federated Vision-Language-Recommendation with Personalized Fusion
Zhiwei Li (University of Technology Sydney), Qiang Yang (Hong Kong Polytechnic University)
Recommendation SystemFederated LearningTransformerMixture of ExpertsVision Language ModelMultimodality
🎯 What it does: Propose FedVLR under the federated learning framework to address the user personalization fusion problem in vision-language recommendation
FedGRPO: Privately Optimizing Foundation Models with Group-Relative Rewards from Domain Clients
Gongxi Zhu, Yuxing Han (Tsinghua University)
OptimizationFederated LearningSafty and PrivacyTransformerLarge Language ModelReinforcement LearningTextBenchmark
🎯 What it does: Propose the FedGRPO framework, converting large model optimization into a reward-based RL evaluation process, combining expert selection and group relative reward aggregation to achieve federated learning.
FedLAGC: Towards High Performance System-Heterogeneous Federated Learning via Layer-Adaptive Submodel Extraction and Gradient Correction
Qing Hu, Chuan Chen (Sun Yat-Sen University)
Federated LearningConvolutional Neural NetworkImage
🎯 What it does: Propose the FedLAGC framework, achieving resource adaptation and communication efficiency in system heterogeneous federated learning through layer-adaptive submodel extraction and gradient correction.
FedMerge: Federated Model Merging for Personalization
Shutong Chen, Chengqi Zhang (University Of Technology Sydney)
Federated LearningImageText
🎯 What it does: Propose the FedMerge framework, which uses the server side to perform weighted merging of multiple models, generating a single personalized model for each client, rather than having clients download/train multiple models.
FedP²EFT: Federated Learning to Personalize PEFT for Multilingual LLMs
Royson Lee (Samsung AI Center), Timothy Hospedales (Samsung AI Center)
Federated LearningTransformerLarge Language ModelSupervised Fine-TuningText
🎯 What it does: Propose FedP EFT, which jointly learns personalized parameter-efficient fine-tuning (LoRA) structures for multilingual LLMs in cross-device federated learning environments, achieving adaptive LoRA rank allocation for each client.
FedPKDA: Personalized Federated Learning with Privacy-Preserving Knowledge Dynamic Alignment
Moxuan Zeng (Hainan University), Jieren Cheng (Hainan University)
Federated LearningSafty and PrivacyImage
🎯 What it does: This paper proposes the FedPKDA framework, aiming to achieve privacy-preserving personalized federated learning and enhance the generalization ability of local models on heterogeneous data through dynamic knowledge alignment.
FedPM: Federated Learning Using Second-order Optimization with Preconditioned Mixing of Local Parameters
Hiro Ishii (Institute of Science Tokyo), Rio Yokota (Institute of Science Tokyo)
ClassificationOptimizationFederated LearningConvolutional Neural NetworkImage
🎯 What it does: Designed and verified a federated learning algorithm named FedPM, which achieves global second-order optimization by using preconditioned mixed local parameters on the server side, thereby improving convergence speed and accuracy in heterogeneous data environments.
FedRNC: Addressing Spatio-Temporal Label Misalignment in Federated Noisy Class-Incremental Learning
Xingwei Huang (Huazhong University of Science and Technology), Zengqiang Yan (Huazhong University of Science and Technology)
ClassificationFederated LearningImageBenchmark
🎯 What it does: Propose FedRNC, a two-stage federated denoising framework, to address spatiotemporal label errors (STLM) in federated noisy incremental learning (FNCIL), by first coarsely filtering noise through local loss clustering, and then performing fine-grained label correction in the feature space using global prototypes.
FedSDA: Federated Stain Distribution Alignment for Non-IID Histopathological Image Classification
Cheng-Chang Tsai (Academia Sinica), Chun-Shien Lu (Academia Sinica)
ClassificationDomain AdaptationFederated LearningDiffusion modelAuto EncoderImageBiomedical Data
🎯 What it does: This paper proposes the FedSDA method, which addresses the feature distribution shift problem in non-IID histopathology image classification by aligning staining distributions across clients within a federated learning framework.
FedSDWC: Federated Synergistic Dual-Representation Weak Causal Learning for OOD
Zhenyuan Huang (Beihang University), Haijun Yang (Beihang University)
Domain AdaptationAnomaly DetectionFederated LearningImage
🎯 What it does: Propose the FedSDWC model, which integrates invariant and variant features and leverages weak causal relationships to achieve out-of-distribution (OOD) generalization and detection in federated learning.
FedSEA-LLaMA: A Secure, Efficient and Adaptive Federated Splitting Framework for Large Language Models
Zishuai Zhang (Beihang University), Zhiming Zheng (Beihang University)
Federated LearningSafty and PrivacyComputational EfficiencyTransformerLarge Language ModelSupervised Fine-TuningTextBenchmark
🎯 What it does: Propose FedSEA-LLaMA, a secure, efficient, and adaptable partitioning framework for dividing LLMs between clients and servers in a federated environment.
FedShard: Federated Unlearning with Efficiency Fairness and Performance Fairness
Siyuan Wen (Hong Kong University of Science and Technology), Ningning Ding (Hong Kong University of Science and Technology)
Federated LearningImage
🎯 What it does: Proposed a hierarchical sharding federated learning and unlearning framework named FedShard, which can balance efficiency fairness and performance fairness during the unlearning process.
FedSkeleton: Secure Multi-Party Graph Skeleton Construction for Privacy-Preserving Federated Time-Series Forecasting
Henggang Deng (Tsinghua University), Tao Jiang (Huazhong University of Science and Technology)
Federated LearningSafty and PrivacyGraph Neural NetworkGraphTime SeriesOrdinary Differential Equation
🎯 What it does: Propose the FedSkeleton framework, leveraging graph skeleton construction and dual-stream prediction for privacy-preserving federated time series forecasting
FedTopo: Topology-Informed Representation Alignment in Federated Learning Under Non-I.I.D. Conditions
Ke Hu (Shanghai Jiao Tong University), Weidong Qiu (Shanghai Jiao Tong University)
Federated LearningRepresentation LearningConvolutional Neural NetworkImage
🎯 What it does: Propose the FedTopo framework, achieving cross-client feature alignment in non-I.I.D. federated learning through topological information.
FeTS: A Feature-Aware Framework for Time Series Forecasting
Le Wang (Shenzhen University), Songbai Liu (Shenzhen University)
Computational EfficiencyTransformerTime Series
🎯 What it does: Propose the FeTS framework, which includes two major modules, AdaFE and DSFFN, to address the issue of uneven importance in time series prediction through adaptive feature extraction and dual-scale fusion;
Few-Shot Precise Event Spotting via Unified Multi-Entity Graph and Distillation
Zhaoyu Liu (National University of Singapore), Jin Song Dong (National University of Singapore)
RecognitionObject DetectionPose EstimationKnowledge DistillationRecurrent Neural NetworkGraph Neural NetworkVideoMultimodality
🎯 What it does: Proposed and implemented a unified multi-entity graph network, UMEG-Net, for precise event detection under few-shot conditions.
Few-step Flow for 3D Generation via Marginal-Data Transport Distillation
Zanwei Zhou (Shanghai Jiao Tong University), Qi Tian (Huawei Inc)
GenerationData SynthesisKnowledge DistillationTransformerFlow-based ModelPoint CloudMeshOrdinary Differential Equation
🎯 What it does: Proposed a new 'MDT-dist' framework to achieve minimal-step inference for flow models in 3D generation tasks
FGD-Align: Pluralistic Alignment for Large Language Models via Fuzzy Group Decision-Making
Weihang Pan (Zhejiang University), Jieping Ye (Zhejiang University)
Reinforcement Learning from Human FeedbackLarge Language ModelText
🎯 What it does: Propose the FGD-Align framework, which utilizes fuzzy group decision theory to achieve multi-perspective alignment of large language models, integrating triangular fuzzy numbers, hierarchical aggregation, and probabilistic fuzzy DPO;
FGM-HD: Boosting Generation Diversity of Fractal Generative Models through Hausdorff Dimension Induction
Haowei Zhang (Sichuan University), Mao Li (Sichuan University)
GenerationConvolutional Neural NetworkDiffusion modelImage
🎯 What it does: This paper proposes a FGM-HD framework based on Hausdorff dimension, enhancing the diversity of fractal generative models while maintaining image quality through self-learning HD estimation and dynamic weight scheduling.
FGNet: Leveraging Feature-Guided Attention to Refine SAM2 for 3D EM Neuron Segmentation
Zhenghua Li (Tsinghua University), Xiaolin Hu (Tsinghua University)
SegmentationTransformerBiomedical DataBenchmark
🎯 What it does: Propose the FGNet framework, which transfers the pre-trained Segment Anything 2 (SAM2) visual prior to three-dimensional electron microscopy (EM) neuron segmentation. The framework employs feature-guided attention (FGA) to guide the fine-grained encoder (FGE) in extracting details, and uses dual affinity decoders to generate refined segmentation results.
FIA-Edit: Frequency-Interactive Attention for Efficient and High-Fidelity Inversion-Free Text-Guided Image Editing
Kaixiang Yang (Huazhong University of Science and Technology), Zhiwei Wang (Huazhong University of Science and Technology)
Image TranslationTransformerDiffusion modelRectified FlowImageVideoBiomedical DataBenchmark
🎯 What it does: This paper proposes a non-reversible text-guided image editing framework called FIA-Edit, achieving high fidelity and semantic accuracy through frequency interaction attention;
Fidelity-Aware Recommendation Explanations via Stochastic Path Integration
Oren Barkan (Open University), Noam Koenigstein (Tel Aviv University)
Recommendation SystemExplainability and InterpretabilityTabular
🎯 What it does: Proposed SPINRec, a stochastic path integral explanation method for recommendation systems, utilizing random baseline sampling to capture the effects of observed and unobserved interactions;
FilmSceneDesigner: Chaining Set Design for Procedural Film Scene Generation
Zhifeng Xie (Shanghai University), Mengtian Li (Shanghai University)
GenerationTransformerAgentic AITextMeshRetrieval-Augmented GenerationChain-of-Thought
🎯 What it does: Propose an automated movie scene generation system called FilmSceneDesigner, which simulates professional scene design workflows. It generates structured parameters based on natural language descriptions and completes tasks such as wall/column structures, material assignments, window/door installations, and prop layouts through a programmatic pipeline, ultimately producing complete movie scenes.
FilmWeaver: Weaving Consistent Multi-Shot Videos with Cache-Guided Autoregressive Diffusion
Xiangyang Luo (Tsinghua University), Shao-Lun Huang (Kuaishou Technology)
GenerationData SynthesisVision Language ModelDiffusion modelVideoTextRetrieval-Augmented Generation
🎯 What it does: Propose the FilmWeaver framework, which achieves long-term and short-term consistency in multi-shot video generation of arbitrary length using autoregressive diffusion models.
Filter, Correlate, Compress: Training-Free Token Reduction for MLLM Acceleration
Yuhang Han (Westlake University), Siteng Huang (Zhejiang University)
CompressionComputational EfficiencyTransformerLarge Language ModelImageVideoTextMultimodalityBenchmark
🎯 What it does: Proposed a three-stage 'Filter-Correlate-Compress' framework to reduce multimodal token counts in visual encoders and LLM decoders without training, significantly enhancing large model inference speed;
FILTER: A Framework for Defending Against Backdoor Attacks in Vertical Federated Learning
Zhanyi Hu (East China Normal University), Yanhao Wang (East China Normal University)
Federated LearningSafty and PrivacyAdversarial AttackImageTabular
🎯 What it does: Proposed the FILTER framework to defend against backdoor attacks in vertical federated learning (VFL), ensuring model integrity during training.
FIND: A Simple Yet Effective Baseline for Diffusion-Generated Image Detection
Jie Li, Jiayi Ji (Xiamen University)
ClassificationAnomaly DetectionConvolutional Neural NetworkTransformerDiffusion modelImageBenchmark
🎯 What it does: Propose a model-free image generation detection method called FIND, which trains a binary classifier by adding noise to real images and labeling them as synthetic images, thereby capturing the distribution differences between real and diffusion-generated images.
Finding Diverse Solutions Parameterized by Cliquewidth
Karolina Drabik (University of Warsaw), Tomáš Masařík (University of Warsaw)
OptimizationComputational EfficiencyGraph
🎯 What it does: This paper proposes a generic framework that transforms dynamic programming (DP) for 'single' vertex problems solvable individually on a given tree-like wide decomposition into DP for diversity problems, achieving diversity solutions under Clique-Width parameterization, and further introduces a new Venn diversity measure;
Finding One Local Optimum Is Easy—but What About Two?
Yasuaki Kobayashi (Hokkaido University), Yutaro Yamaguchi (Osaka University)
OptimizationGraph
🎯 What it does: This paper studies the complexity of finding multiple local optima in unweighted combinatorial optimization problems, particularly proving that computing two local optima is NP-hard in various natural unweighted local search problems.
Finding the Translation Switch: Discovering and Exploiting the Task-Initiation Features in LLMs
Xinwei Wu (Tianjin University), Kaifu Zhang (Alibaba International Digital Commerce)
Explainability and InterpretabilityComputational EfficiencyRepresentation LearningData-Centric LearningTransformerLarge Language ModelSupervised Fine-TuningAuto EncoderText
🎯 What it does: This paper utilizes sparse autoencoders (SAE) to identify and validate features in large language models (LLMs) responsible for translation initiation, and applies them to efficient data selection and fine-tuning;
Finding Time Series Anomalies Using Granular-Ball Vector Data Description
Lifeng Shen (Chongqing University of Posts and Telecommunications), Yi Liu (Chongqing Ant Consumer Finance Co Ltd)
Anomaly DetectionRecurrent Neural NetworkTime SeriesBenchmark
🎯 What it does: Proposed the Granular-ball One-Class Network (GBOC), which maps time-series data into an adaptive high-density particle ball space via Granular-ball Vector Data Description, combined with an LSTM encoder for unsupervised anomaly detection.