AAAI 2025 Papers — Page 16
AAAI Conference on Artificial Intelligence · 3028 papers
Integrating Personalized Spatio-Temporal Clustering for Next POI Recommendation
Chao Song (University of Electronic Science and Technology of China), Li Lu (University of Electronic Science and Technology of China)
Recommendation SystemGraph Neural NetworkTransformerTime SeriesSequential
🎯 What it does: This paper proposes a next location recommendation model integrating personalized spatiotemporal clustering, iPCM, to predict the user's next visit point in location-based services.
Integrating Sequence and Image Modeling in Irregular Medical Time Series Through Self-Supervised Learning
Liuqing Chen (Zhejiang University), Lingyun Sun (Zhejiang University)
ClassificationRepresentation LearningRecurrent Neural NetworkTransformerGenerative Adversarial NetworkContrastive LearningTime SeriesBiomedical Data
🎯 What it does: A joint learning framework is proposed that simultaneously uses sequence representation and image representation for the classification of irregular medical time series, and three self-supervised learning strategies are designed to fuse the two representations.
Intelligent OPC Engineer Assistant for Semiconductor Manufacturing
Guojin Chen (Chinese University of Hong Kong), Haoxing Ren (NVIDIA)
OptimizationReinforcement Learning from Human FeedbackLarge Language ModelReinforcement LearningMultimodality
🎯 What it does: This paper proposes a two-stage framework that first uses reinforcement learning (PPO) to automatically explore OPC recipes for different layouts, and then utilizes a multimodal LLM (GPT-4o) for feature extraction, decision tree construction, and recipe induction based on the RL results, ultimately generating recipe rules that can be directly used in commercial OPC software.
Intent Oriented Contrastive Learning for Sequential Recommendation
Wuhong Wang (University of Science and Technology of China), Zheng Zhang (University of Science and Technology of China)
Recommendation SystemTransformerContrastive LearningSequential
🎯 What it does: This paper proposes a sequence recommendation framework IOCLRec that captures user dynamic intentions using subsequence partitioning and three contrastive learning modules (fine-grained, single intention, multiple intentions).
Interacted Object Grounding in Spatio-Temporal Human-Object Interactions
Xiaoyang Liu (Shanghai Jiao Tong University), Yong-Lu Li (Shanghai Jiao Tong University)
Object DetectionSegmentationTransformerVision Language ModelVideoPoint CloudBenchmark
🎯 What it does: A new open-world human-object interaction video benchmark GIO is proposed, and an object localization task is defined on it. Subsequently, a 4D-QA framework based on SAM candidates and 4D question answering is introduced to address this task.
Interpretable Face Anti-Spoofing: Enhancing Generalization with Multimodal Large Language Models
Guosheng Zhang (Baidu Inc), Jingdong Wang (Baidu Inc)
ClassificationRecognitionExplainability and InterpretabilityTransformerLarge Language ModelVision Language ModelImageMultimodality
🎯 What it does: An interpretable facial anti-spoofing method I-FAS is proposed, transforming the traditional binary classification task into a visual question answering (VQA) framework.
Interpretable Failure Detection with Human-Level Concepts
Kien X. Nguyen (University of Delaware), Xi Peng (University of Delaware)
Anomaly DetectionExplainability and InterpretabilityTransformerLarge Language ModelVision Language ModelImage
🎯 What it does: A post-processing failure detection framework based on human visual concepts, ORCA, is proposed, which uses concept-activated ordinal ranking to estimate model confidence and can explain the reasons for errors.
Interpretable Solutions for Multi-Physics PDEs Using T-NNGP
Lulu Cao (Xiamen University), Min Jiang (Xiamen University)
OptimizationExplainability and InterpretabilityRecurrent Neural NetworkReinforcement LearningTime SeriesPhysics Related
🎯 What it does: A T-NNGP method is proposed, which combines traditional numerical solving with deep reinforcement learning in a hybrid genetic programming approach to automatically generate interpretable symbolic expressions that satisfy multi-physics PDE systems.
Interweaving Memories of a Siamese Large Language Model
Xin Song (East China Normal University), Wei Lu (Wuhan University)
ClassificationGenerationTransformerLarge Language ModelSupervised Fine-TuningText
🎯 What it does: This paper proposes a model-agnostic parameter-efficient fine-tuning framework called IMSM, which enhances the adaptability to downstream tasks by constructing a dual-tower (siamese) LLM and interweaving the memories of the two towers, while alleviating catastrophic forgetting.
Intra and Inter Parser-Prompted Transformers for Effective Image Restoration
Cong Wang (Shenzhen Campus of Sun Yat-sen University), Wei Wang (Dalian University of Technology)
RestorationTransformerImage
🎯 What it does: A PPTformer model is proposed, which generates parsing information based on the visual foundation model (SAM) to guide image restoration, integrating a Transformer structure with parsing prompts.
Intrinsic Image Decomposition for Robust Self-supervised Monocular Depth Estimation on Reflective Surfaces
Wonhyeok Choi (Daegu Gyeongbuk Institute of Science and Technology), Sunghoon Im (Daegu Gyeongbuk Institute of Science and Technology)
Depth EstimationKnowledge DistillationConvolutional Neural NetworkImage
🎯 What it does: A self-supervised monocular depth estimation framework is proposed, which identifies and excludes reflective areas through intrinsic image decomposition, thereby improving depth prediction accuracy.
Inverse Reinforcement Learning by Estimating Expertise of Demonstrators
Mark Beliaev (University of California Santa Barbara), Ramtin Pedarsani (University of California Santa Barbara)
Reinforcement LearningTabular
🎯 What it does: A new inverse reinforcement learning framework IRLEED is proposed to learn true rewards and policies from suboptimal and heterogeneous demonstrations.
InverseCoder: Self-improving Instruction-Tuned Code LLMs with Inverse-Instruct
Yutong Wu (Institute of Computing Technology), Yunji Chen
AI Code AssistantTransformerLarge Language ModelSupervised Fine-TuningPrompt EngineeringText
🎯 What it does: A self-supervised data augmentation method called Inverse-Instruct is proposed, which expands the instruction-tuning dataset for code LLMs by transforming existing code snippets into new natural language instructions and pairing them with the original code, resulting in the training of the InverseCoder series models.
Invertible Projection and Conditional Alignment for Multi-Source Blended-Target Domain Adaptation
Yuwu Lu (South China Normal University), Xue Hu (South China Normal University)
Domain AdaptationFlow-based ModelImage
🎯 What it does: This paper proposes an Inverse Projective and Conditional Alignment method (IPCA) for Multi-Source Blended Domain Adaptation (MBDA). It maps source domain and blended target domain features through reversible projection, enhances model robustness using projection consistency regularization and conditional entropy maximization, and achieves domain-invariant feature learning through CKB metric-driven unsupervised adversarial learning.
Investigating Relational State Abstraction in Collaborative MARL
Sharlin Utke (University of Warwick), Giovanni Montana (University of Warwick)
Robotic IntelligenceGraph Neural NetworkReinforcement LearningGraph
🎯 What it does: This paper proposes a Minimization of Abstract State based on Relative Spatial Relationships (MARC), which transforms multi-agent observations into a graph structure and uses R-GCN for encoding, achieving cooperative learning within a centralized training and distributed execution actor-critic framework.
Investigating the Security Threat Arising from “Yes-No” Implicit Bias in Large Language Models
Yanrui Du (Harbin Institute of Technology), Bing Qin (Harbin Institute of Technology)
Adversarial AttackTransformerLarge Language ModelPrompt EngineeringText
🎯 What it does: This paper quantifies and verifies the existence of implicit bias in 'yes-no' responses by statistically analyzing the probability differences in large language models (LLMs) when answering 'yes' and 'no'. It proposes a covert jailbreak attack method based on this bias (IB-ICM), which induces the model to convert malicious instructions that should be rejected into affirmative responses by inserting high 'yes' inclination instructions into the attack context.
InvSeg: Test-Time Prompt Inversion for Semantic Segmentation
Jiayi Lin (Sony AI), Shaogang Gong (Queen Mary University of London)
SegmentationDiffusion modelContrastive LearningImage
🎯 What it does: During testing, a pre-trained text-image diffusion model is used to invert image-specific visual context into the text prompt embedding space, resulting in region-level custom prompts for each image, thereby achieving more complete and accurate semantic segmentation.
IOP: An Idempotent-Like Optimization Method on the Pareto Front of Hypernetwork
Hui Wang (Beihang University), Xudong Liu (Beihang University)
Domain AdaptationOptimizationMultimodality
🎯 What it does: A new Idempotent-like Optimization (IOP) method is proposed for Pareto Hypernetwork (PHN) to directly generate a complete Pareto front from raw data samples, significantly enhancing adaptability to distribution shifts.
IPDN: Image-enhanced Prompt Decoding Network for 3D Referring Expression Segmentation
Qi Chen (Xiamen University), Xiaoshuai Sun (Xiamen University)
SegmentationTransformerPrompt EngineeringImagePoint Cloud
🎯 What it does: This paper proposes the Image-enhanced Prompt Decoding Network (IPDN), which enhances 3D point cloud semantics through multi-view image information and utilizes task-driven prompts to guide the decoder to focus on the target object, thereby achieving 3D Referring Expression Segmentation.
IPVTON: Image-based 3D Virtual Try-on with Image Prompt Adapter
Xiaojing Zhong (South China University of Technology), Qingyao Wu (South China University of Technology)
GenerationData SynthesisPrompt EngineeringDiffusion modelScore-based ModelImage
🎯 What it does: This paper presents IPVTON, an image-based 3D virtual fitting framework that utilizes IP-Adapter, Score Distillation Sampling, and pseudo-contour loss to achieve high-quality try-ons of 3D human bodies and clothing.
IRMamba: Pixel Difference Mamba with Layer Restoration for Infrared Small Target Detection
Mingjin Zhang (Xidian University), Jie Guo (Xidian University)
Object DetectionConvolutional Neural NetworkImage
🎯 What it does: A new infrared small target detection network called IRMamba based on Mamba is proposed, which includes two main modules: Pixel Difference Mamba (PDMamba) and Layer Restoration Module (LRM).
Is LLMs Hallucination Usable? LLM-based Negative Reasoning for Fake News Detection
Chaowei Zhang (Yangzhou University), Yun Li (The Education University of Hong Kong)
ClassificationKnowledge DistillationTransformerLarge Language ModelText
🎯 What it does: Using LLM hallucination to generate positive and negative reasoning to assist in fake news detection
Is Sarcasm Detection a Step-by-Step Reasoning Process in Large Language Models?
Ben Yao (University of Copenhagen), Jing Qin (Hong Kong Polytechnic University)
TransformerLarge Language ModelPrompt EngineeringTextChain-of-Thought
🎯 What it does: Proposes the SarcasmCue framework, integrating four prompt sub-methods (CoC, GoC, BoC, ToC) to enable large language models to reason step-by-step and concurrently synthesize multiple cues in sarcasm detection.
Is There No Such Thing as a Bad Question? H4R: HalluciBot for Ratiocination, Rewriting, Ranking, and Routing
William Watson (J.P. Morgan AI Research), Nishan Srishankar (J.P. Morgan AI Research)
OptimizationComputational EfficiencyTransformerLarge Language ModelReinforcement LearningText
🎯 What it does: This paper proposes HalluciBot, an encoder model that predicts whether a query is likely to generate hallucinations during the pre-inference stage without calling LLMs; based on this, it implements the H4R framework for query rewriting, ranking, and routing.
Is Your Image a Good Storyteller?
Xiujie Song (Shanghai Jiao Tong University), Kenny Q. Zhu (University of Texas at Arlington)
TransformerLarge Language ModelVision Language ModelImageChain-of-Thought
🎯 What it does: This paper proposes the Image Semantic Assessment (ISA) task, constructs the first ISA dataset containing two scoring metrics: entity complexity and semantic complexity, and designs the VLISA method based on visual language models.
ISPDiffuser: Learning RAW-to-sRGB Mappings with Texture-Aware Diffusion Models and Histogram-Guided Color Consistency
Yang Ren (Sichuan University), Shuaicheng Liu (University of Electronic Science and Technology of China)
Image TranslationRestorationDiffusion modelImage
🎯 What it does: The ISPDiffuser framework is proposed, which splits the RAW-to-sRGB task into two steps: grayscale detail reconstruction and color consistency mapping, achieving high-quality imaging through a texture-aware diffusion model and a histogram-guided color consistency module.
ISR-DPO: Aligning Large Multimodal Models for Videos by Iterative Self-Retrospective DPO
Daechul Ahn (Seoul National University), Jonghyun Choi (Seoul National University)
OptimizationReinforcement Learning from Human FeedbackTransformerReinforcement LearningVision Language ModelVideoTextMultimodality
🎯 What it does: This paper proposes an Iterative Self-Backtracking Direct Preference Optimization (ISR-DPO) method aimed at video, which uses self-generated visual context to guide the preference selection of video-text models, enhancing video question-answering performance.
IsUMap: Manifold Learning and Data Visualization Leveraging Vietoris-Rips Filtrations
Parvaneh Joharinad (Max Planck Institute for Mathematics in the Sciences), Jürgen Jost (Max Planck Institute for Mathematics in the Sciences)
Biomedical Data
🎯 What it does: A hybrid manifold learning algorithm named IsUMap is proposed, which integrates UMAP and Isomap, and utilizes Vietoris-Rips persistence and fuzzy simplicial metrics to achieve low-dimensional embedding.
Iterative Counterfactual Data Augmentation
Mitchell Plyler (North Carolina State University), Min Chi (North Carolina State University)
Explainability and InterpretabilityData-Centric LearningTransformerLarge Language ModelText
🎯 What it does: A method for Iterative Counterfactual Data Augmentation (ICDA) has been designed and implemented to improve interpretability and reduce bias in text classification tasks.
Iterative Self-Training with Class-Aware Text-to-Image Synthesis for Visual Task Learning
Xiang Zhang (Northwest University), Jianping Fan (Northwest University)
Object DetectionSegmentationGenerationData SynthesisConvolutional Neural NetworkLarge Language ModelDiffusion modelImage
🎯 What it does: This paper proposes the IST-CATS framework, which utilizes class-aware text-to-image synthesis to generate diverse and annotated synthetic images. It trains semantic segmentation and object detection models using both synthetic and unlabeled real images through iterative self-training and the LabFilt strategy.
Iterative Sparse Attention for Long-sequence Recommendation
Guanyu Lin (Carnegie Mellon University), Depeng Jin (Tsinghua University)
Recommendation SystemTransformerSequential
🎯 What it does: This paper proposes an ISA model for long sequence recommendation, which combines sparse attention layers and iterative attention layers to achieve efficient modeling of long behavior sequences and capture interest drift.
IteRPrimE: Zero-shot Referring Image Segmentation with Iterative Grad-CAM Refinement and Primary Word Emphasis
Yuji Wang (Shenzhen International Graduate School Tsinghua University), Yansong Tang (Shenzhen International Graduate School Tsinghua University)
Object DetectionSegmentationTransformerVision Language ModelImage
🎯 What it does: A zero-shot referential image segmentation framework called IteRPrimE is proposed, which completes segmentation by aligning the target regions in the original image using Grad-CAM.
ITP: Instance-Aware Test Pruning for Out-of-Distribution Detection
Haonan Xu (Nanjing University of Science and Technology), Yang Yang (Pazhou Lab Guangzhou)
Anomaly DetectionConvolutional Neural NetworkImage
🎯 What it does: Proposed and validated the Instance-aware Test Pruning (ITP) method for post-processing on pre-trained models to enhance OOD detection performance.
IWRN:A Robust Blind Watermarking Method for Artwork Image Copyright Protection Against Noise Attack
Feifei Kou (Beijing University of Posts and Telecommunications), Xuejing Kang (Beijing University of Posts and Telecommunications)
Data SynthesisSafty and PrivacyConvolutional Neural NetworkContrastive LearningImage
🎯 What it does: A robust blind watermarking method for artistic images, IWRN, is proposed, which can resist various noise attacks while maintaining invisibility.
JAQ: Joint Efficient Architecture Design and Low-Bit Quantization with Hardware-Software Co-Exploration
Mingzi Wang (Tsinghua University), Wenwu Zhu (Tsinghua University)
OptimizationComputational EfficiencyNeural Architecture SearchImage
🎯 What it does: Jointly designing neural network structures, ultra-low bit mixed precision quantization, and dedicated accelerator architectures to construct an end-to-end differentiable JAQ framework;
JEN-1 Composer: A Unified Framework for High-Fidelity Multi-Track Music Generation
Yao Yao (Jen Music AI), Alex Wang (Jen Music AI)
GenerationData SynthesisDiffusion modelAudio
🎯 What it does: This paper proposes the JEN-1 Composer, which integrates a single audio latent diffusion model to unify the modeling of the edge, conditional, and joint distributions of multi-track music, achieving controllable high-fidelity multi-track music generation.
JEN-1 DreamStyler: Customized Musical Concept Learning via Pivotal Parameters Tuning
Boyu Chen (Jen Music AI), Alex Wang (Jen Music AI)
GenerationData SynthesisTransformerDiffusion modelAudio
🎯 What it does: A custom text-to-music generation method named JEN-1 DreamStyler is proposed, which learns from two minutes of reference music to generate new music that embodies the concept of that music (such as instruments or genres).
Joint Class-level and Instance-level Relationship Modeling for Novel Class Discovery
Jiaying Zhou (Peking University), Qingchao Chen (Peking University)
ClassificationRecognitionConvolutional Neural NetworkImage
🎯 What it does: A new method is proposed for simultaneously modeling class-level and instance-level relationships and knowledge transfer in Novel Class Discovery (NCD).
Joint Knowledge Editing for Information Enrichment and Probability Promotion
Wenhang Shi (Renmin University of China), Xiaoyong Du
TransformerContrastive LearningText
🎯 What it does: A knowledge editing method called JEEP is proposed, which simultaneously edits low-level information richness and high-level probability enhancement, utilizing contrastive detection to identify key stages of editing.
JoVALE: Detecting Human Actions in Video Using Audiovisual and Language Contexts
Taein Son (Hanyang University), Jun Won Choi (Seoul National University)
RecognitionTransformerVision Language ModelVideoMultimodalityAudio
🎯 What it does: This paper proposes a multi-modal video action detection framework called JoVALE, which combines audio, visual, and language contextual information to achieve action recognition through actor-centric feature aggregation.
Just a Few Glances: Open-Set Visual Perception with Image Prompt Paradigm
Jinrong Zhang (Dalian University of Technology), Zhengnan Hu (Xiaomi AI Lab)
Object DetectionSegmentationTransformerPrompt EngineeringContrastive LearningImage
🎯 What it does: Proposes the Image Prompt Paradigm and MI Grounding framework, utilizing a small number of image instances as prompts to achieve open-set object detection and segmentation.
Just What You Desire: Constrained Timeline Summarization with Self-Reflection for Enhanced Relevance
Muhammad Reza Qorib (National University of Singapore), Hwee Tou Ng (National University of Singapore)
TransformerLarge Language ModelPrompt EngineeringText
🎯 What it does: This paper proposes the Constrained Timeline Summarization (CTLS) task and constructs a new dataset called CREST. It also designs a training-free method based on large language models (LLM) called REACTS, which generates constraint-compliant timelines through steps such as self-reflection, clustering, and summarization.
K-hop Hypergraph Neural Network: A Comprehensive Aggregation Approach
Linhuang Xie (Xiamen University), Taisong Jin (South China Normal University)
Graph Neural NetworkGraph
🎯 What it does: A K-hop hypergraph neural network (KHGNN) based on path features is proposed to address the issues of over-compression and over-smoothing in traditional hypergraph GNNs.
k-HyperEdge Medoids for Clustering Ensemble
Feijiang Li (Institute of Big Data Science and Industry Shanxi University), Liang Du (Institute of Big Data Science and Industry Shanxi University)
OptimizationTabular
🎯 What it does: A clustering ensemble method based on k-HyperEdge Medoids, CEHM, is proposed. It first constructs the base clustering results into a hypergraph, and then iteratively obtains the final k non-overlapping hyperedges through three steps: initialization, diffusion, and tuning, directly providing the clustering results.
K-ON: Stacking Knowledge on the Head Layer of Large Language Model
Lingbing Guo (Zhejiang University), Huajun Chen (Zhejiang University)
TransformerLarge Language ModelContrastive LearningMultimodality
🎯 What it does: In the task of knowledge graph completion, the author improves alignment effectiveness by stacking multi-head predictions in the head layer of a large language model to generate entities in one go and introducing entity-level contrastive learning.
KAES: Multi-aspect Shared Knowledge Finding and Aligning for Cross-prompt Automated Scoring of Essay Traits
Xia Li (Guangdong University of Foreign Studies), Wenjing Pan (Guangdong University of Foreign Studies)
OptimizationConvolutional Neural NetworkRecurrent Neural NetworkTransformerLarge Language ModelText
🎯 What it does: An optimized framework for multi-dimensional shared knowledge discovery and alignment, KAES, is proposed for cross-prompt automated essay scoring.
KALAHash: Knowledge-Anchored Low-Resource Adaptation for Deep Hashing
Shu Zhao (Pennsylvania State University), Vijaykrishnan Narayanan (Pennsylvania State University)
RetrievalDomain AdaptationVision Language ModelImage
🎯 What it does: Under extremely low resources (such as 1-shot, 2-shot), adaptive learning is applied to pre-trained deep hashing models to address the decline in retrieval performance caused by distribution shift.
KDAT: Inherent Adversarial Robustness via Knowledge Distillation with Adversarial Tuning for Object Detection Models
Yarin Yerushalmi Levi (Ben-Gurion University of the Negev), Yuval Elovici (Ben-Gurion University of the Negev)
Object DetectionKnowledge DistillationAdversarial AttackImage
🎯 What it does: To address the robustness of target detection models against adversarial patch attacks, a KDAT mechanism is proposed, which enhances the detection performance of the model on both adversarial and normal images through knowledge distillation and adversarial fine-tuning while maintaining the original inference time.
Kernel Learning for Sample Constrained Black-Box Optimization
Rajalaxmi Rajagopalan (University of Illinois), Romit Roy Choudhury (University of Illinois)
Recommendation SystemOptimizationAuto EncoderImageTime SeriesAudio
🎯 What it does: This paper proposes a kernel learning framework called KOBO based on variational autoencoders, aimed at automatically learning the optimal kernel in black-box optimization with limited samples;
Kernel-Aware Graph Prompt Learning for Few-Shot Anomaly Detection
Fenfang Tao (Nanjing University of Science and Technology), Xiangbo Shu (Nanjing University)
Anomaly DetectionGraph Neural NetworkPrompt EngineeringImage
🎯 What it does: This paper proposes a few-shot anomaly detection framework based on Kernel-Aware Hierarchical Graph (KAG-prompt).
KernelMatmul: Scaling Gaussian Processes to Large Time Series
Tilman Hoffbauer (RWTH Aachen University), Jakob Bossek (Paderborn University)
Time Series
🎯 What it does: Proposes KernelMatmul, which accelerates Gaussian process regression and reduces memory usage for large-scale irregular time series.
Key-Point-Driven Data Synthesis with Its Enhancement on Mathematical Reasoning
Yiming Huang (Microsoft), Weizhu Chen (Microsoft)
GenerationData SynthesisTransformerLarge Language ModelSupervised Fine-TuningText
🎯 What it does: A keypoint-driven data synthesis method (KPDDS) is proposed, generating large-scale mathematical reasoning datasets KPMath and KPMath-Plus. Subsequently, various LLMs are fine-tuned under supervision, improving performance on multiple mathematical benchmarks.
KeyGS: A Keyframe-Centric Gaussian Splatting Method for Monocular Image Sequences
Keng-Wei Chang (National Tsing Hua University), Shang-Hong Lai (National Tsing Hua University)
Pose EstimationOptimizationGaussian SplattingSimultaneous Localization and MappingImageVideo
🎯 What it does: Proposes the KeyGS framework, which uses keyframe SfM to quickly obtain rough camera poses and jointly optimizes with 3D Gaussian Splatting to achieve high-quality 3D reconstruction.
KeyPose: Category-Level 6D Object Pose Estimation with Self-Adaptive Keypoints
Sheng Yu (Beijing Institute of Technology), Yuanqing Xia (Zhongyuan University of Technology)
Pose EstimationGraph Neural NetworkTransformerPoint Cloud
🎯 What it does: A category-level 6D pose estimation method based on adaptive keypoints, KeyPose, is proposed, which predicts keypoints using a Transformer and achieves precise localization in point clouds.
KGCRR: An Effective Metric-Driven Knowledge Graph Completion Framework by Designing a Novel Upper Bound Function with Adaptive Approximation to Reciprocal Rank
Kuan Xu (Beijing Jiaotong University), Xuezhong Zhou (Beijing Jiaotong University)
Recommendation SystemOptimizationGraph Neural NetworkGraph
🎯 What it does: A framework named KGCRR is proposed, which effectively optimizes the reciprocal ranking (RR) metric in knowledge graph completion by designing a new upper bound function CRR, addressing the inconsistency between existing loss functions and the RR objective.
KITS: Inductive Spatio-Temporal Kriging with Increment Training Strategy
Qianxiong Xu (Nanyang Technological University), Zhishuai Li (SenseTime Research)
Graph Neural NetworkGraphTime Series
🎯 What it does: An incremental training strategy named KITS is proposed, which uses virtual nodes to simulate unobserved nodes during training. This approach combines reference node feature fusion and node-aware cyclic adjustment to perform semi-supervised learning on spatial-temporal Kriging, addressing the 'graph gap' issue between training and inference graphs.
Knockoffs Inference for Partially Linear Models with Automatic Structure Discovery
Hao Wang (Huazhong Agricultural University), Hong Chen (Huazhong Agricultural University)
Tabular
🎯 What it does: A variable selection method for partially linear models based on Knockoffs, called KI-LAND, is proposed, achieving FDR control for both linear and nonlinear variables and automatic structure discovery.
Know Where You Are From: Event-Based Segmentation via Spatio-Temporal Propagation
Ke Li (Beijing University of Technology), Yongjian Deng (Beijing University of Technology)
SegmentationAutonomous DrivingTransformerImageVideo
🎯 What it does: A semantic segmentation framework KWYAF based on event cameras is proposed, which enhances segmentation accuracy by utilizing historical motion information through spatiotemporal propagation.
Know2Vec: A Black-Box Proxy for Neural Network Retrieval
Zhuoyi Shang (Institute of Information Engineering Chinese Academy of Sciences), Xiangyang Ji (Tsinghua University)
ClassificationRetrievalRecurrent Neural NetworkTabular
🎯 What it does: This paper proposes Know2Vec, a method for model retrieval in black-box environments through knowledge consistency matching.
Knowledge Editing with Dynamic Knowledge Graphs for Multi-Hop Question Answering
Yifan Lu (Harbin Institute of Technology), Min Zhang (Harbin Institute of Technology)
Graph Neural NetworkGraphRetrieval-Augmented GenerationChain-of-Thought
🎯 What it does: A knowledge editing method based on dynamic knowledge graphs, KEDKG, is proposed to enhance the accuracy and reliability of multi-hop question answering.
Knowledge Graph Completion with Relation-Aware Anchor Enhancement
Duanyang Yuan (National University of Defense Technology), Jian Huang (National University of Defense Technology)
Graph Neural NetworkContrastive LearningGraph
🎯 What it does: A knowledge graph completion method called RAA-KGC is proposed, which utilizes relation-aware anchor enhancement;
Knowledge in Superposition: Unveiling the Failures of Lifelong Knowledge Editing for Large Language Models
Chenhui Hu (Institute of Automation, Chinese Academy of Sciences), Jun Zhao (Institute of Automation, Chinese Academy of Sciences)
TransformerLarge Language ModelText
🎯 What it does: This paper discusses the failure mechanisms of large language models (LLMs) in lifelong knowledge editing, and theoretically derives and experimentally verifies that knowledge superposition is the fundamental cause of cumulative interference in editing, ultimately leading to model forgetting.
Knowledge Is Power: Harnessing Large Language Models for Enhanced Cognitive Diagnosis
Zhiang Dong (Zhejiang University), Fei Wu (Zhejiang University)
Large Language ModelAuto EncoderContrastive LearningText
🎯 What it does: A model-agnostic KCD framework is proposed, utilizing large language models to enhance knowledge in cognitive diagnostic models and align the semantic-behavioral space.
Knowledge-Enhanced Hierarchical Heterogeneous Graph for Personality Identification with Limited Training Data
Yuxuan Song (Chinese Academy of Sciences), Daniel Dajun Zeng (City University of Hong Kong)
RecognitionGraph Neural NetworkContrastive LearningText
🎯 What it does: This paper proposes a knowledge-enhanced hierarchical heterogeneous graph model to achieve personalized recognition with a small amount of labeled data.
Knowledge-Guided Domain Adaptation Model for Transferring Drug Response Prediction from Cell Lines to Patients
Xuan Liu (Xinyang Normal University), Menglu Li (Huazhong Agricultural University)
Domain AdaptationDrug DiscoveryGraph Neural NetworkAuto EncoderBiomedical Data
🎯 What it does: A knowledge-driven domain adaptation model, TransDRP, is proposed to transfer drug response prediction from cell lines to patients, supporting one-time predictions for multiple drugs.
KnowPO: Knowledge-Aware Preference Optimization for Controllable Knowledge Selection in Retrieval-Augmented Language Models
Ruizhe Zhang (Peking University), Yasha Wang (Peking University)
RetrievalOptimizationTransformerTextRetrieval-Augmented Generation
🎯 What it does: This paper proposes the KnowPO method, which enhances the knowledge selection ability of retrieval-augmented language models in the face of internal and external knowledge conflicts by constructing a knowledge conflict dataset and preference optimization.
KOALA: Kernel Coupling and Element Imputation Induced Multi-View Clustering
Tingting Wu (Beijing Jiaotong University), Songhe Feng (Beijing Jiaotong University)
OptimizationMultimodality
🎯 What it does: This paper proposes a decoupled missing multi-view clustering method named KOALA, which simultaneously accomplishes kernel matrix reconstruction, view alignment, and clustering through cross-kernel alignment learning and low-rank tensor constraints.
Kolmogorov-Arnold Networks Still Catastrophically Forget but Differently from MLP
Anton Lee (Victoria University of Wellington), W. Bastiaan Kleijn (University of Waikato)
Time SeriesPhysics Related
🎯 What it does: This study verifies that Kolmogorov-Arnold Networks (KAN) still experience catastrophic forgetting in more complex tasks with overlapping inputs, and proposes a parameter isolation and hierarchical pruning method called WiseKAN for KAN.
KPL: Training-Free Medical Knowledge Mining of Vision-Language Models
Jiaxiang Liu (Zhejiang University), Zuozhu Liu (Zhejiang University)
ClassificationRetrievalTransformerLarge Language ModelVision Language ModelContrastive LearningImageMultimodalityBiomedical Data
🎯 What it does: This study proposes the Knowledge Proxy Learning (KPL) method, which constructs a knowledge-enhanced foundation, utilizes visual retrieval to filter relevant descriptions to generate richer text proxies, and combines the Stable Green Horn algorithm for multimodal proxy learning, significantly improving performance in zero-shot medical image classification tasks.
L-Man: A Large Multi-modal Model Unifying Human-centric Tasks
Jialong Zuo (Huazhong University of Science and Technology), Kai Han (Huawei Noah's Ark Lab)
RecognitionGenerationRetrievalTransformerLarge Language ModelVision Language ModelContrastive LearningImageTextMultimodality
🎯 What it does: A large-scale language-image instruction-following dataset called HumanIns has been constructed, and the L-Man model has been proposed to unify human-centered tasks under open-ended instructions.
L3TC: Leveraging RWKV for Learned Lossless Low-Complexity Text Compression
Junxuan Zhang (Ant Group), Li Song (Shanghai Jiao Tong University)
CompressionRecurrent Neural NetworkLarge Language ModelText
🎯 What it does: This paper proposes a low-complexity lossless text compression method called L3TC, which utilizes the RWKV model to predict probabilities and implements compression using arithmetic coding.
L4DR: LiDAR-4DRadar Fusion for Weather-Robust 3D Object Detection
Xun Huang (Xiamen University), Cheng Wang (Xiamen University)
Object DetectionAutonomous DrivingPoint Cloud
🎯 What it does: This paper proposes a 3D object detection framework L4DR that integrates LiDAR and 4D radar, aiming to enhance detection robustness in adverse weather conditions.
Label Aggregation for Composite Crowd Tasks by Worker Ability Constraint Satisfaction
Jiyi Li (University of Yamanashi)
Text
🎯 What it does: This paper proposes a label aggregation method for Composite Crowd Tasks (CCT), which bridges worker information across different tasks by satisfying worker capability constraints, enabling multi-task joint aggregation.
Label Noise Correction via Fuzzy Learning Machine
Jiye Liang (Shanxi University), Junbiao Cui (Shanxi University)
ClassificationAnomaly DetectionSupervised Fine-TuningImage
🎯 What it does: This paper proposes a two-stage label noise correction framework (FLM-LNC) based on Fuzzy Learning Machine (FLM), which detects and corrects noisy samples through fuzzy similarity and tolerable loss.
Label-Efficient Data Augmentation with Video Diffusion Models for Guidewire Segmentation in Cardiac Fluoroscopy
Shaoyan Pan (Emory University), Shanhui Sun (United Imaging Intelligence)
SegmentationGenerationData SynthesisDiffusion modelVideo
🎯 What it does: Using the two-stage video synthesis framework SF-VD based on diffusion models, combined with scene and motion modeling, as well as frame-consistent sampling and segmentation guidance, to generate data suitable for guiding line segmentation.
Label-Free Backdoor Attacks in Vertical Federated Learning
Wei Shen (Wuhan University), Mang Ye (Wuhan University)
Federated LearningAdversarial AttackImageTabular
🎯 What it does: A backdoor attack method without label information is proposed in vertical federated learning, utilizing embedded gradients to construct poisoned samples of the target class, and enhancing the attack effect through selective sample exchange.
LAGD: Local Topological-Alignment and Global Semantic-Deconstruction for Incremental 3D Semantic Segmentation
Yumin Zhang (Beijing Jiaotong University), Bo Wei (Newcastle University)
SegmentationKnowledge DistillationPoint Cloud
🎯 What it does: This paper proposes an incremental semantic segmentation framework for 3D point clouds called LAGD, which can avoid catastrophic forgetting when new categories are continuously added.
LAMA-UT: Language Agnostic Multilingual ASR Through Orthography Unification and Language-Specific Transliteration
Sangmin Lee (Yonsei University), Hong-Goo Kang (Yonsei University)
RecognitionTransformerLarge Language ModelPrompt EngineeringAudio
🎯 What it does: This paper proposes a multilingual speech recognition pipeline LAMA-UT without language-specific modules: it first generates a unified romanized transcription using wav2vec2.0-XLSR, and then converts this transcription into the original text of various languages using a frozen large language model (LLM).
Langevin Monte Carlo Beyond Lipschitz Gradient Continuity
Matej Benko (Brno University of Technology), Błażej Miasojedow (University of Warsaw)
OptimizationImageStochastic Differential Equation
🎯 What it does: This paper proposes the Inexact Proximal Langevin Algorithm (IPLA), which extends the sampling capability of Langevin Monte Carlo (LMC) to non L-smooth, super-quadratic potential by using approximate proximal operations in Wasserstein space, and provides corresponding error analysis.
Langevin Multiplicative Weights Update with Applications in Polynomial Portfolio Management
Yi Feng (Shanghai University of Finance and Economics), Tian Xie (Shanghai University of Finance and Economics)
OptimizationTime SeriesFinance RelatedStochastic Differential Equation
🎯 What it does: This paper proposes a new algorithm for global non-convex optimization under multi-dimensional simplex constraints—Langevin Multiplier Weight Update (LMWU)—and proves that it can converge to an internal global optimum at a non-asymptotic rate under certain conditions.
Language Model Can Listen While Speaking
Ziyang Ma (Shanghai Jiao Tong University), Xie Chen (ByteDance Inc.)
TransformerLarge Language ModelAudio
🎯 What it does: An end-to-end full-duplex interactive speech language model (LSLM) has been developed, enabling simultaneous listening and speaking in real-time speech environments, and supporting instant interruption in human-computer dialogue.
Language Models of Code Are Few-Shot Planners and Reasoners for Multi-Document Summarization with Attribution
Abhilash Nandy (Indian Institute of Technology Kharagpur), Sambaran Bandyopadhyay (Adobe Research)
GenerationAI Code AssistantGraph Neural NetworkTransformerLarge Language ModelPrompt EngineeringText
🎯 What it does: This paper proposes the MiDAS-PRo framework, which uses Code-LLM for planning and reasoning in multi-document summarization, incorporating source citations in the generated summaries to address the issues of chaotic structure and missing attribution in multi-source summaries.
Language Pre-training Guided Masking Representation Learning for Time Series Classification
Liaoyuan Tang (Northwestern Polytechnical University), Feiping Nie (Northwestern Polytechnical University)
ClassificationRepresentation LearningAuto EncoderContrastive LearningTime Series
🎯 What it does: This paper proposes a self-supervised time series representation learning framework guided by language pre-training (LPMRL), which enhances the discriminability and generalization ability of time series by adaptively selecting semantic blocks for masking through language embeddings, combined with autoencoder reconstruction and dual information contrastive learning.
Language Prompt for Autonomous Driving
Dongming Wu (Beijing Institute of Technology), Jianbing Shen (University of Macau)
Object TrackingAutonomous DrivingTransformerPrompt EngineeringVideoText
🎯 What it does: The NuPrompt dataset is proposed, and based on this, the PromptTrack model is developed to achieve language prompt-driven 3D multi-object tracking and trajectory prediction.
Large Images Are Gaussians: High-Quality Large Image Representation with Levels of 2D Gaussian Splatting
Lingting Zhu (University of Hong Kong), Lequan Yu (University of Hong Kong)
GenerationOptimizationGaussian SplattingImage
🎯 What it does: For large-sized images, a LIG framework based on 2D Gaussian Splatting (2DGS) is proposed, utilizing a variant representation that allows direct optimization of the covariance matrix and a two-level Gaussian hierarchy (Level-of-Gaussian) to achieve high-quality fitting.
Large Language Model Meets Graph Neural Network in Knowledge Distillation
Shengxiang Hu (Shanghai University), Yixin Chen (Washington University in St. Louis)
Knowledge DistillationGraph Neural NetworkTransformerLarge Language ModelContrastive LearningTextGraph
🎯 What it does: The LinguGKD framework is proposed, which distills the semantic knowledge of large language models into graph neural networks, achieving efficient learning of text attribute graphs.
Large Language Models Are Read/Write Policy-Makers for Simultaneous Generation
Shoutao Guo (Chinese Academy of Sciences), Yang Feng (Chinese Academy of Sciences)
GenerationTransformerLarge Language ModelPrompt EngineeringTextMultimodalityAudio
🎯 What it does: A synchronous generation framework LSG based on large language models is proposed, allowing the LLM to read input in real-time while deciding the timing of generation and outputting text.
Large Language Models Enhanced Personalized Graph Neural Architecture Search in Federated Learning
Hui Fang (Zhejiang University), Haishuai Wang (Zhejiang University)
Federated LearningNeural Architecture SearchGraph Neural NetworkLarge Language ModelPrompt EngineeringGraph
🎯 What it does: In a federated learning environment, a personalized graph neural network architecture search is achieved by combining large language models and hypernetworks, automatically generating GNN architectures that adapt to the heterogeneous graph data of each client.
Last-iterate Convergence in Regularized Graphon Mean Field Game
Jing Dong (Chinese University of Hong Kong), Yaoliang Yu (University of Waterloo)
OptimizationReinforcement Learning from Human FeedbackGraph Neural NetworkReinforcement LearningGraph
🎯 What it does: The convergence rate of the last iteration using the mirror descent algorithm in graphical mean field games is proposed and analyzed, proving that the convergence rate under bandit feedback is O(T^{-1/4}), under full information is O(T^{-1}), and in linear GMFG achieves a convergence rate of O(T^{-1/5});
Latent Diffusion-Enhanced Virtual Try-On via Optimized Pseudo-Label Generation
Chenghu Du (Wuhan University of Technology), Shengwu Xiong (Shanghai Artificial Intelligence Laboratory)
GenerationKnowledge DistillationDiffusion modelFlow-based ModelImage
🎯 What it does: A virtual try-on network based on latent diffusion models is proposed, achieving fully supervised learning through cycle consistency and knowledge distillation, addressing the issue of sensitivity to mask errors in self-supervised filling paradigms.
Latent Reward: LLM-Empowered Credit Assignment in Episodic Reinforcement Learning
Yun Qu (Tsinghua University), Xiangyang Ji (Tsinghua University)
TransformerLarge Language ModelReinforcement LearningPrompt EngineeringSequential
🎯 What it does: This paper proposes a Latent Reward framework (LaRe) based on large language models (LLMs), which generates multi-dimensional and interpretable task performance metrics and achieves finer credit allocation through return decomposition.
LATTE: Improving Latex Recognition for Tables and Formulae with Iterative Refinement
Nan Jiang (Purdue University), Lin Tan (Purdue University)
RecognitionImage TranslationTransformerSupervised Fine-TuningVision Language ModelImageTabular
🎯 What it does: A LaTeX recognition framework called LATTE is proposed, which can generate renderable LaTeX source code from images of tables and formulas and continuously improve.
LayerAct: Advanced Activation Mechanism for Robust Inference of CNNs
Kihyuk Yoon (Ulsan National Institute of Science and Technology), Chiehyeon Lim (Ulsan National Institute of Science and Technology)
ClassificationSegmentationAdversarial AttackConvolutional Neural NetworkImageBiomedical Data
🎯 What it does: This paper proposes LayerAct—a mechanism for activation in CNNs achieved through layer normalization to enhance the network's robustness to noise;
Layered-Parameter Perturbation for Zeroth-Order Optimization of Optical Neural Networks
Hiroshi Sawada (Communication Science Laboratories, NTT Corporation), Masaya Notomi (Basic Research Laboratories, NTT Corporation)
OptimizationImage
🎯 What it does: This paper proposes a zero-order optimization method for optical neural networks—layered parameter perturbation, which utilizes a specially designed covariance matrix to perturb module parameters, thereby making the output perturbation more isotropic and improving the black-box training effect.
LazyDiT: Lazy Learning for the Acceleration of Diffusion Transformers
Xuan Shen (Middle Tennessee State University), Jiuxiang Gu (Adobe Research)
GenerationData SynthesisComputational EfficiencyTransformerDiffusion modelImage
🎯 What it does: The LazyDiT framework is proposed, which utilizes cached results and linear layers to predict similarity at each time step of the diffusion Transformer, thereby dynamically skipping redundant computations to achieve inference acceleration.
LCGC: Learning from Consistency Gradient Conflicting for Class-Imbalanced Semi-Supervised Debiasing
Weiwei Xing (Beijing Jiaotong University), Xinyu Pang (Chongqing University of Posts and Telecommunications)
ClassificationContrastive LearningImage
🎯 What it does: The LCGC method is proposed, which improves classifier bias in class-imbalanced semi-supervised learning through consistency gradient conflict learning and debiasing with reference images.
LDP: Generalizing to Multilingual Visual Information Extraction by Language Decoupled Pretraining
Huawen Shen (Institute of Information Engineering), Yu Zhou (Nankai University)
RecognitionSegmentationDomain AdaptationTransformerDiffusion modelImageText
🎯 What it does: In this work, the authors propose a new multilingual visual information extraction pre-training paradigm called LDP, and based on this, they construct the LDM model, which can achieve strong cross-language generalization capabilities with pre-training using only English data.
Learn How to Query from Unlabeled Data Streams in Federated Learning
Yuchang Sun (Hong Kong University of Science and Technology), Jun Zhang (Hong Kong University of Science and Technology)
Federated LearningConvolutional Neural NetworkTransformerReinforcement LearningImageText
🎯 What it does: A multi-agent reinforcement learning method for querying unlabeled streaming data in federated learning (LeaDQ) is proposed, enabling clients to make sample labeling decisions based on the global model state without sharing data.
LEARN: Knowledge Adaptation from Large Language Model to Recommendation for Practical Industrial Application
Jian Jia (Kuaishou Technology), Kun Gai (Kuaishou Technology)
Recommendation SystemTransformerLarge Language ModelContrastive LearningTextSequential
🎯 What it does: Construct the LEARN framework to transfer the open-world knowledge of large language models (LLMs) into recommendation systems, and encode user history and item text through a dual-tower structure (user tower and item tower) to obtain user/item vectors that can be directly used for online ranking.
Learn2Aggregate: Supervised Generation of Chvatal-Gomory Cuts Using Graph Neural Networks
Arnaud Deza (University of Toronto), Yong Zhang (Huawei Technologies Canada)
OptimizationGraph Neural NetworkGraph
🎯 What it does: By training a graph neural network for binary classification of MILP constraints, the focus is placed only on important constraints during CG cutting generation, significantly reducing the size of the separation problem.