AAAI 2026 Papers — Page 6
AAAI Conference on Artificial Intelligence · 4149 papers
Causal-ERC: A Multimodal Framework with Causal Prompting for Emotion Recognition in Conversations with Large Language Models
Ran Jing (Harbin Institute of Technology), Ruifeng Xu (Harbin Institute of Technology)
RecognitionRecurrent Neural NetworkTransformerLarge Language ModelPrompt EngineeringMultimodalityBenchmark
🎯 What it does: Proposed the Causal-ERC framework, leveraging large language models and multimodal fusion with causal prompts for dialogue emotion recognition.
Causal-HalBench: Uncovering LVLMs Object Hallucinations Through Causal Intervention
Zhe Xu (University of Science and Technology of China), Xiang Wang (University of Science and Technology of China)
RestorationVision Language ModelImageTextMultimodalityBenchmark
🎯 What it does: This paper constructs a causal analysis framework, proposes Visual Content Intervention (VCI) to automatically generate counterfactual samples, and creates the Causal-HalBench benchmark for systematically evaluating object hallucination issues in LVLMs.
Causal-Tune: Mining Causal Factors from Vision Foundation Models for Domain Generalized Semantic Segmentation
Yin Zhang (Harbin Institute of Technology), Dan Liu (Universitat Autonoma De Barcelona)
SegmentationDomain AdaptationAutonomous DrivingTransformerImage
🎯 What it does: In domain generalization semantic segmentation tasks, by performing frequency domain (DCT) decomposition on Vision Foundation Models, causal and non-causal features are extracted, and causal features are optimized using learnable causal markers.
Causal, Strategic, and Combined Responsibility Attribution in Situation Calculus Concurrent Game Structures
MohammadHossein Karimian, Yves Lesperance (York University)
🎯 What it does: Under the synchronous situation calculus game structure (SCSGS) framework, this paper proposes a formal responsibility attribution method that combines causal responsibility with strategic responsibility, and introduces a causal chain based on minimal action subsets to precisely identify actual causal contributions.
CausalCLIP: Causally-Informed Feature Disentanglement and Filtering for Generalizable Detection of Generated Images
Bo Liu (Chongqing University of Posts and Telecommunications), Qinghui He (Chongqing University of Posts and Telecommunications)
Anomaly DetectionTransformerVision Language ModelContrastive LearningImage
🎯 What it does: Designed a causal representation-based image detection framework called CausalCLIP, which first separates CLIP features into causal and non-causal subspaces, then filters out non-causal features through adversarial masks and HSIC constraints to achieve cross-model general detection of generated images.
Causality Matters: How Temporal Information Emerges in Video Language Models
Yumeng Shi (Nanyang Technological University), Wenya Wang (Nanyang Technological University)
Computational EfficiencyVision Language ModelVideoMultimodalityBenchmark
🎯 What it does: This paper investigates the temporal understanding mechanisms in video language models (VideoLM), revealing that they do not rely on positional encoding but instead generate sequential sensitivity through causal attention; meanwhile, it proposes two efficient inference strategies based on this mechanism;
Causality-Aligned Semantic Recovery for Incomplete Cross-Modal Retrieval
Haipeng Chen (Jilin University), Yingda Lyu (University of Science and Technology of China)
RetrievalGraph Neural NetworkLarge Language ModelVision Language ModelImageVideoText
🎯 What it does: Proposes the CASR method for recovering missing modalities and eliminating pseudo-relevance in visual-text retrieval tasks, significantly improving the performance of incomplete modality retrieval.
Causality-Aware Efficient Exploration for Cooperative Multi-Agent Reinforcement Learning
Hongye Cao, Yang Gao (University of Liverpool)
Computational EfficiencyReinforcement LearningBenchmark
🎯 What it does: Proposed an efficient exploration framework called CEE based on causal relationships to improve sample efficiency in cooperative multi-agent reinforcement learning.
Causality-inspired Federated Learning for Dynamic Spatio-Temporal Graphs
Yuxuan Liu (University of Electronic Science and Technology of China), Boyuan Zhang (University of Electronic Science and Technology of China)
Federated LearningGraph Neural NetworkContrastive LearningGraphTime Series
🎯 What it does: This paper proposes a federated learning framework named SC-FSGL based on causal separation for distributed prediction on dynamic spatiotemporal graphs.
Causally-Aware Attribute Completion for Incomplete Federated Graph Clustering
Jingxin Liu (Hainan University), Xiangyan Tang (Hainan University)
Federated LearningGraph Neural NetworkGraph
🎯 What it does: This study proposes IFedGC, a framework capable of performing node-level clustering in federated graph clustering scenarios with missing attributes;
Causally-Grounded Dual-Path Attention Intervention for Object Hallucination Mitigation in LVLMs
Liu Yu (University of Electronic Science and Technology of China), Gillian Dobbie (University of Electronic Science and Technology of China)
Explainability and InterpretabilityTransformerVision Language ModelContrastive LearningMultimodality
🎯 What it does: Propose the Owl framework by causal modeling of visual and textual attention, using fine-grained attention intervention and dual-path contrastive decoding to reduce object misreporting in large vision-language models.
CausalStep: A Benchmark for Explicit Stepwise Causal Reasoning in Videos
Xuchen Li (Institute of Automation, Chinese Academy of Sciences), Wentao Zhang (Zhongguancun Academy)
Large Language ModelVideoBenchmarkChain-of-Thought
🎯 What it does: Proposes CausalStep, a video causal reasoning benchmark, employing manual causal segmentation, strict stepwise QA processes, and interference options based on error types, aiming to rigorously evaluate models' step-by-step causal reasoning capabilities.
CauVQ: Causal Vector Quantization for Graph OOD Generalization
Weihong Zhang (Shanxi University), Xian Yang (The University of Manchester)
Domain AdaptationExplainability and InterpretabilityGraph Neural NetworkGraph
🎯 What it does: Proposes a graph learning framework based on causal vector quantization (CauVQ), which enhances the generalization and interpretability of graph neural networks in out-of-distribution (OOD) environments by mapping local substructures to discrete codebooks and leveraging causal discovery with counterfactual regularization.
CCAHCL: Multi-Level Hypergraph Contrastive Learning for Connected Component Awareness
Zhuo Li (Beijing University of Technology), Zun Li (Beijing University of Technology)
Representation LearningGraph Neural NetworkContrastive LearningGraph
🎯 What it does: Proposes a connected component-aware hypergraph contrastive learning framework named CCAHCL, which constructs multi-layer representations for nodes, hyperedges, and connected components, and balances the learning of different-scale connected components through hierarchical contrastive loss.
CCFQA: A Benchmark for Cross-Lingual and Cross-Modal Speech and Text Factuality Evaluation
Yexing Du (Harbin Institute of Technology), Yang Xiang (Pengcheng Laboratory)
TransformerLarge Language ModelSupervised Fine-TuningTextMultimodalityBenchmarkAudio
🎯 What it does: Proposed CCFQA, a cross-lingual and cross-modal (speech and text) factuality evaluation benchmark, and used it to systematically assess the fact consistency of existing multimodal large language models.
CD-DPE: Dual-Prompt Expert Network Based on Convolutional Dictionary Feature Decoupling for Multi-Contrast MRI Super-Resolution
Xianming Gu (Guizhou University), Yi Chen (Guizhou University)
Super ResolutionConvolutional Neural NetworkMixture of ExpertsBiomedical DataMagnetic Resonance Imaging
🎯 What it does: Achieve multi-contrast MRI super-resolution reconstruction using a dual-prompt expert network and convolutional dictionary feature decoupling method, precisely leveraging structural information from high-resolution reference images to guide the restoration of low-resolution target images.
CEC-Zero: Zero-Supervision Character Error Correction with Self-Generated Rewards
Zhiming Lin (Nankai University), Canran Xiao (Shanghai High School International Division)
TransformerLarge Language ModelReinforcement LearningText
🎯 What it does: Construct a zero-supervised reinforcement learning framework that enables large language models to correct Chinese spelling errors through self-generated errors and self-rewarding mechanisms.
CellStream: Dynamical Optimal Transport Informed Embeddings for Reconstructing Cellular Trajectories from Snapshots Data
Yue Ling (Peking University), Peijie Zhou (Peking University)
Auto EncoderBiomedical DataOrdinary Differential Equation
🎯 What it does: Propose the CellStream framework, which jointly learns low-dimensional embeddings and cell dynamics to recover continuous cell trajectories from single-cell snapshot data.
Center-Outward q-Dominance: A Sample-Computable Proxy for Strong Stochastic Dominance in Stochastic Multi-Objective Optimisation
Robin van der Laag (Leiden University), Yingjie Fan (Leiden University)
OptimizationTabularBenchmark
🎯 What it does: Propose the center-outward q-advantage relationship as a computable proxy for strong first-order stochastic dominance, validated in two scenarios: hyperparameter optimization and noisy multi-objective evolutionary algorithms.
Centralized Group Equitability and Individual Envy-Freeness in the Allocation of Indivisible Items
Ying Wang (Columbia University), Minming Li (City University of Hong Kong)
OptimizationComputational Efficiency
🎯 What it does: Studied the fair allocation of divisible items to pre-defined groups under the perspective of a centralized allocator, while considering both individual and centralized fairness.
Certified Branch-and-Bound MaxSAT Solving
Dieter Vandesande (Vrije Universiteit Brussel), Bart Bogaerts (Universitat de Girona)
OptimizationExplainability and InterpretabilityBenchmark
🎯 What it does: Added verifiable proof logging functionality to the branch-and-bound MaxSAT solver MAXCDCL, enabling evidence generation for look-ahead bound estimation and MDD encoding.
Certified but Fooled! Breaking Certified Defenses with Ghost Certificates
Quoc Viet Vo (University of Adelaide), Damith C. Ranasinghe (Monash University)
ClassificationSegmentationAdversarial AttackImage
🎯 What it does: Propose a region-aware adversarial attack called GhostCert, which can deceive provably robust models based on randomized smoothing by generating false large robustness radius (ghost certificate) while maintaining input semantics unchanged.
Certified L2-Norm Robustness of 3D Point Cloud Recognition in the Frequency Domain
Liang Zhou (Hohai University), Tianze Chen (Hohai University)
RecognitionGraph Neural NetworkPoint Cloud
🎯 What it does: This paper proposes FreqCert, a robustness certification framework for 3D point cloud classification based on frequency domain subsampling and voting.
CertMask: Certifiable Defense Against Adversarial Patches via Theoretically Optimal Mask Coverage
Xuntao Lyu (North Carolina State University), Zhishan Guo (North Carolina State University)
ClassificationAdversarial AttackImage
🎯 What it does: Proposed CertMask, a single-round deterministic mask coverage certifiable robust defense to counter adversarial patch attacks.
CGMIS: Concept-Graph Based Multi-Hop Instructions Synthesis for Enhancing Long-Context Reasoning
Zechen Sun (Soochow University), Qiaoming Zhu (Soochow University)
Data SynthesisLarge Language ModelTextGraphChain-of-Thought
🎯 What it does: Proposes the CGMIS framework, which explicitly constructs multi-hop reasoning paths using concept graphs to automatically generate verifiable multi-hop instruction data.
Chain-of-Search: Parameter-Efficient Reasoning for Zero-Shot Object Navigation
Hanrui Chen (Hangzhou Dianzi University), Pan Li (Meta AI)
Autonomous DrivingRobotic IntelligenceTransformerPrompt EngineeringVision-Language-Action ModelImageText
🎯 What it does: Designed the Chain-of-Search (CoS) framework to address zero-shot target navigation tasks while avoiding the use of large language models;
Channel-masked Asymmetric Distribution Matching for Cross-Domain Generalized Dataset Distillation
Qi Liu (Xidian University), Yanhua Yang (Xidian University)
Domain AdaptationKnowledge DistillationGenerative Adversarial NetworkImage
🎯 What it does: Propose a Channel Mask Asymmetric Distribution Matching (CADM) framework for cross-domain generalizable dataset distillation.
CharBench: Evaluating the Role of Tokenization in Character-Level Tasks
Omri Uzan (Stanford University), Yuval Pinter (Ben-Gurion University of the Negev)
TransformerLarge Language ModelTextBenchmark
🎯 What it does: Constructed a benchmark framework named CHARBENCH containing 175,000 examples with two character-level tasks (counting and localization), systematically evaluating character reasoning capabilities of multiple public and proprietary large language models.
CHARM: Collaborative Harmonization Across Arbitrary Modalities for Modality-Agnostic Semantic Segmentation
Lekang Wen (State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University), Mi Wang (Hangzhou Institute of Technology, Xidian University)
SegmentationTransformerMultimodality
🎯 What it does: Proposes the CHARM framework, aiming to achieve cross-modal synergy and harmonization, addressing the issue of mode homogenization in multi-modal semantic segmentation.
ChartEditor: A Reinforcement Learning Framework for Robust Chart Editing
Liangyu Chen (Renmin University of China), Qin Jin (Renmin University of China)
TransformerLarge Language ModelSupervised Fine-TuningReinforcement LearningVision Language ModelMultimodalityBenchmark
🎯 What it does: Proposes the ChartEditVista benchmark and ChartEditor model to address the no-code chart editing task.
CHASE: Contextual History for Adaptive and Simple Exploitation in Large Language Model Jailbreaking
Zhiqiang Hao (Nanjing University), Vincent Ng (University of Texas at Dallas)
Adversarial AttackLarge Language ModelPrompt EngineeringTextBenchmark
🎯 What it does: Proposes CHASE, a jailbreak method based on multi-turn dialogue, which first generates 'jailbroken history' on vulnerable LLMs and then transfers this history to target LLMs, inducing them to output harmful content while maintaining conversational coherence.
CHDP: Cooperative Hybrid Diffusion Policies for Reinforcement Learning in Parameterized Action Space
Bingyi Liu (Wuhan University of Technology), Zhuangzhuang Zhang (City University of Hong Kong)
Reinforcement LearningDiffusion modelAuto EncoderBenchmark
🎯 What it does: Proposes the Cooperative Hybrid Diffusion Policies (CHDP) framework, which generates discrete and continuous actions through two-agent diffusion policies in parameterized action spaces, and achieves scalable and expressive decision-making via sequence updates and Q-guided codebooks.
Cheating Stereo Matching in Full-Scale: Physical Adversarial Attack Against Binocular Depth Estimation in Autonomous Driving
Kangqiao Zhao (Nanyang Technological University), Jun Luo (Nanyang Technological University)
Depth EstimationAutonomous DrivingAdversarial AttackImageMesh
🎯 What it does: This paper investigates the safety of stereo matching depth estimation models in autonomous driving, proposing a physically deployable full-surface 3D texture adversarial sample capable of deceiving binocular depth estimation under varying perspectives, lighting, and weather conditions;
CHIMERA: Controllable High-quality Image-Mask Extraction for Reliable Diffusion-based Anomaly Synthesis
JoungBin Lee (KAIST AI), Seungryong Kim (KAIST AI)
GenerationData SynthesisAnomaly DetectionTransformerVision Language ModelDiffusion modelImageTextMultimodality
🎯 What it does: This paper proposes the CHIMERA framework, which can synthesize high-quality, controllable, and generalizable anomaly samples in industrial images based on natural language instructions.
Chinese Two-part Allegorical Sayings Reading Comprehension: Exploration from Reasoning to Metaphor
Dongyu Su (Key Laboratory of Smart Farming for Agricultural Animals), Ying Sha (Huazhong Agricultural University)
TransformerLarge Language ModelContrastive LearningTextBenchmark
🎯 What it does: Constructed the Chinese Two-Fable Idiom (TPAS) Reading Comprehension Dataset (CTRC) and proposed a Multi-Perspective TPAS Contrastive Learning Network (MTCLN) to address three major challenges: rhetoric identification, logical reasoning, and metaphor understanding.
ChipMind: Retrieval-Augmented Reasoning for Long-Context Circuit Design Specifications
Changwen Xing (Southeast University), Jun Yang (Southeast University)
TransformerLarge Language ModelTextGraphBenchmarkRetrieval-Augmented GenerationChain-of-Thought
🎯 What it does: Built a retrieval-augmented reasoning framework called ChipMind based on a knowledge graph to address multi-hop reasoning problems in long-text IC specifications.
CIA: Cluster-Instance Alignment for Unsupervised Day-Night Vehicle Re-Identification
Yongguo Ling (Guangxi University), Wenhao Shao (Guangxi University)
RecognitionDomain AdaptationConvolutional Neural NetworkContrastive LearningImage
🎯 What it does: Propose a fully unsupervised day-night vehicle re-identification framework (CIA) that addresses illumination domain shift by solving the problem through cluster and instance dual-level alignment.
CiNuSeg: Class Incremental Nuclei Segmentation via Anchor-driven Consistency Learning with Dual Region Regularization
Xuexin Wu (Shenzhen University), Jing Qin (Hong Kong Polytechnic University)
SegmentationKnowledge DistillationConvolutional Neural NetworkBiomedical Data
🎯 What it does: Proposes the CiNuSeg method to achieve incremental nuclear segmentation, addressing the balance between old class forgetting and new class learning.
CIP-Net: Continual Interpretable Prototype-based Network
Federico Di Valerio (Sapienza University), Roberto Capobianco (Sony AI)
ClassificationExplainability and InterpretabilityComputational EfficiencyContrastive LearningImage
🎯 What it does: Propose a sample-free self-explaining prototype network called CIP-Net, achieving continual learning through a shared prototype layer.
Circuit-Think: A Multimodal Reasoning Framework for Automated Circuit-to-Netlist Translation with Trajectory-Guided Reinforcement Learning
Yuqi Jiang (Zhejiang University), Cheng Zhuo (Zhejiang University)
Image TranslationLarge Language ModelSupervised Fine-TuningReinforcement LearningVision Language ModelMultimodalityChain-of-Thought
🎯 What it does: Propose the Circuit-Think framework for the automatic translation task from circuit images to SPICE netlists.
City Sampling for Citizens’ Assemblies
Paul Gölz, Philipp C. Verpoort (Sortition Foundation)
OptimizationTabular
🎯 What it does: This paper studies the two-stage sampling problem when holding citizens' assemblies in countries such as Germany, and designs an algorithm that ensures equal sampling probabilities for all citizens while limiting the number of participating cities, the number of letters, and upper limits under municipal registry constraints.
CKDA: Cross-modality Knowledge Disentanglement and Alignment for Visible-Infrared Lifelong Person Re-identification
Zhenyu Cui (Peking University), Yuxin Peng (Peking University)
RecognitionRetrievalDomain AdaptationKnowledge DistillationRepresentation LearningTransformerPrompt EngineeringContrastive LearningImageMultimodality
🎯 What it does: Propose a cross-modal knowledge separation and alignment method (CKDA), which separates the cross-modal shared knowledge and the modality-specific knowledge exclusive to visible/infrared modalities through visual prompting, and achieves dual alignment of new and old knowledge without data replay, thereby mitigating catastrophic forgetting in visible-infrared lifelong person re-identification;
CL-DMDF: Dynamic Multimodal Data Fusion Model Based on Contrastive Learning
Dong Li (Liaoning University), Yue Kou (Northeastern University)
SegmentationRepresentation LearningContrastive LearningMultimodality
🎯 What it does: Propose a contrastive learning-based dynamic multi-modal data fusion model (CL-DMDF), aiming to achieve efficient and adaptive multi-modal information fusion in scenarios with missing or noisy modalities;
CL-Guard: Defending DNNs Against Backdoors via Fine-Grained Neuron Analysis and Collaborative Dual-Network Learning
Jie Xiao (Zhejiang University of Technology), Fan Terry Zhang (Zhejiang University of Technology)
Safty and PrivacyKnowledge DistillationAdversarial AttackConvolutional Neural NetworkImage
🎯 What it does: Proposes the CL-Guard pre-deployment defense framework, which selects critical neurons through recursive hierarchical partitioning, sparsely trains non-critical neurons, and introduces dual-network collaborative learning to eliminate backdoors in deep neural networks.
CLASP: Cross-modal Salient Anchor-based Semantic Propagation for Weakly-supervised Dense Audio-Visual Event Localization
Jinxing Zhou (Mohamed Bin Zayed University of Artificial Intelligence), Dan Guo (National University of Singapore)
Object DetectionTransformerContrastive LearningVideoMultimodalityAudio
🎯 What it does: Investigated the problem of dense audio-visual event localization under weak supervision, proposing the CLASP method based on cross-modal salient anchor points for semantic propagation;
Class Incremental Medical Image Segmentation via Prototype-Guided Calibration and Dual-Aligned Distillation
Shengqian Zhu (Sichuan University), Junjie Hu (Sichuan University)
SegmentationKnowledge DistillationConvolutional Neural NetworkImageBiomedical DataComputed Tomography
🎯 What it does: Two methods, prototype-based calibrated distillation (PGCD) and bidirectional aligned prototype distillation (DAPD), are proposed to address the category incremental segmentation problem in medical imaging.
Class-Aware Active Annotation in Federated Semi-Supervised Learning for Medical Image Classification
Meiting Xue (Hangzhou Dianzi University), Jing Ma (Zhejiang Chinese Medical University)
ClassificationFederated LearningImageBiomedical Data
🎯 What it does: In medical image classification tasks, a framework is proposed that combines active learning with federated semi-supervised learning, enabling dynamic updates of the labeled set, improvement of pseudo-label quality, and class balance under non-IID environments.
Class-feature Watermark: A Resilient Black-box Watermark Against Model Extraction Attacks
Yaxin Xiao (Hong Kong Polytechnic University), Haibo Hu (Hong Kong Polytechnic University)
Safty and PrivacyRepresentation LearningAdversarial AttackConvolutional Neural NetworkContrastive LearningImageTextAudio
🎯 What it does: This paper systematically evaluates the security of black-box model watermarks, proposing a novel removal attack method WRK that can break the protection of traditional representation-coupled watermarks. Subsequently, Class-Feature Watermark (CFW) is designed by modifying watermarks into class-level forged samples and combining representation coupling with stability optimization, achieving robust protection against model extraction and removal attacks.
Class-Partitioned VQ-VAE and Latent Flow Matching for Point Cloud Scene Generation
Dasith de Silva Edirimuni (University of Western Australia), Ajmal Saeed Mian (University of Western Australia)
GenerationData SynthesisFlow-based ModelAuto EncoderPoint Cloud
🎯 What it does: Propose a pure point cloud scene generation framework that directly generates object bounding boxes, categories, and latent features using class-partitioned VQ-VAE and latent space flow matching models, avoiding external database retrieval.
Classifier-induced Reciprocal Points for Multi-label Open-set Recognition
Yibo Wang, Min-Ling Zhang (Lenovo Group Ltd)
ClassificationRecognitionTabular
🎯 What it does: Proposes the CREM (Classifier-Induced Reciprocal Points for Multi-Label Open-Set Recognition) framework, combining classifier-induced complementary points, RBF kernel mapping, label relevance matrix, and joint optimization to address the dual tasks of known label classification and unknown label identification in multi-label open-set recognition.
Clean-Label Physical Backdoor Attacks with Data Distillation
Thinh Dao (VinUniversity), Kok-Seng Wong (VinUniversity)
Knowledge DistillationAdversarial AttackImage
🎯 What it does: This paper proposes a Clean Label Physical Backdoor Attack (CLPBA), which implants a backdoor by applying tiny, invisible perturbations to a small number of target class samples, without modifying labels or injecting triggers.
Clear Nights Ahead: Towards Multi-Weather Nighttime Image Restoration
Yuetong Liu (Chongqing University of Posts and Telecommunications), Bin Xiao (Chongqing University of Posts and Telecommunications)
RestorationData SynthesisTransformerMixture of ExpertsImage
🎯 What it does: Proposed and implemented a multi-weather nighttime image restoration task, constructed the AllWeatherNight dataset, and introduced a unified ClearNight framework;
ClearAIR: A Human-Visual-Perception-Inspired All-in-One Image Restoration
Xu Zhang (Wuhan University), Lefei Zhang (Wuhan University)
RestorationTransformerVision Language ModelImage
🎯 What it does: Propose ClearAIR, a global-local hierarchical image restoration framework based on human visual perception;
CLER: Improving Multimodal Financial Reasoning by Cross-MLLM Error Reflection
Shuangyan Deng (University of Auckland), Jiamou Liu (University of Auckland)
Large Language ModelContrastive LearningMultimodalityFinance RelatedRetrieval-Augmented GenerationChain-of-Thought
🎯 What it does: Propose the CLER framework, which enhances the accuracy of multimodal financial reasoning through the retrieval of cross-model errors and iterative reflection.
CliCARE: Grounding Large Language Models in Clinical Guidelines for Decision Support over Longitudinal Cancer Electronic Health Records
Dongchen Li (Northeastern University), Kun Yu (Northeastern University)
Graph Neural NetworkTransformerLarge Language ModelBiomedical DataElectronic Health RecordsRetrieval-Augmented Generation
🎯 What it does: Proposes the CliCARE framework, which converts long-term cancer electronic health records (EHR) into a temporal knowledge graph (TKG), aligns it with clinical guideline knowledge graphs, and generates clinical summaries and treatment recommendations.
Client-level Active Error Correction in Distributed Learning
Kwang In Kim (Pohang University of Science and Technology)
Federated LearningImageBenchmark
🎯 What it does: Proposes a method for client-level active error correction in distributed learning environments.
CLIP-FTI: Fine-Grained Face Template Inversion via CLIP-Driven Attribute Conditioning
Longchen Dai (Jinan University), Zhihua Xia (Jinan University)
GenerationTransformerGenerative Adversarial NetworkImage
🎯 What it does: Reverse reconstruct leaked facial templates to generate high-quality facial images with fine-grained facial attributes.
CLIP2Pose: Frozen CLIP as Semantic Guide for Domain Adaptive Pose Estimation
Jiawen Li (East China Normal University), Aimin Zhou (East China Normal University)
Pose EstimationDomain AdaptationGraph Neural NetworkTransformerPrompt EngineeringVision Language ModelGenerative Adversarial NetworkImage
🎯 What it does: Propose the CLIP2Pose framework, which achieves unsupervised domain adaptation for pose estimation by leveraging a frozen CLIP model and structure-driven prompts.
CLIPDet3D: Vision-Language Collaborative Distillation for 3D Object Detection
Jiaqi Zhao (China University of Mining and Techology), Qigong Sun (SenseTime Research)
Object DetectionAutonomous DrivingTransformerPrompt EngineeringContrastive LearningImageBenchmark
🎯 What it does: Propose CLIPDet3D, a multi-view 3D object detection framework based on audio-visual collaboration, specifically addressing the long-tailed distribution problem of rare categories.
CLIPPan: Adapting CLIP as a Supervisor for Unsupervised Pansharpening
Lihua Jian (Zhengzhou University), Lihui Chen (Chongqing University)
Super ResolutionPrompt EngineeringVision Language ModelContrastive LearningImageTextMultimodality
🎯 What it does: Propose the CLIPPan framework, using CLIP as a supervisor for unsupervised full-resolution pansharpening, guided by text descriptions to direct the fusion process;
Cliqueformer: Model-Based Optimization with Structured Transformers
Jakub Grudzien Kuba (UC Berkeley), Sergey Levine (UC Berkeley)
OptimizationTransformerTabularBiomedical Data
🎯 What it does: Proposed and implemented a Transformer-based Cliqueformer architecture for offline model-based optimization (MBO), which decomposes black-box objective functions by learning predefined functional graph models (FGMs) to achieve more efficient design generation.
CLM-Access: A Specialized Foundation Model for High-Dimensional Single-Cell ATAC-Seq Analysis
Ziqiang Liu (Hangzhou Institute of Medicine, Chinese Academy of Sciences), Xiaolin Li (Hangzhou Institute of Medicine, Chinese Academy of Sciences)
TransformerLarge Language ModelSupervised Fine-TuningMultimodalityBiomedical Data
🎯 What it does: Designed and trained a Transformer-based foundation model called CLM-Access specifically for scATAC-seq, incorporating steps such as unified preprocessing, patch embedding, and masked reconstruction.
Closer to Biological Mechanism: Drug-Drug Interaction Prediction from the Perspective of Pharmacophore
Mingliang Dou (Taiyuan University of Technology), Fei Guo (Central South University)
Explainability and InterpretabilityDrug DiscoveryGraph Neural NetworkBiomedical Data
🎯 What it does: This paper proposes a pharmacophore-based drug-drug interaction prediction method called PC-DDI.
CloserToMe: A Unified Framework for Accurate and Transferable Latency Prediction Across Heterogeneous Devices
Cheng Tang (University of Science and Technology of China), Xuehai Zhou (University of Science and Technology of China)
Computational EfficiencyMeta LearningGraph Neural NetworkGraphBenchmark
🎯 What it does: Designed a unified latency prediction framework called CloserToMe across heterogeneous devices, leveraging device behavior signatures, hardware capability vectors, and a hardware-operation dialogue module to achieve high-precision, transferable inference latency prediction.
CloudMamba: Grouped Selective State Spaces for Point Cloud Analysis
Kanglin Qu, Yuanhao Sun (Beijing University Of Posts And Telecommunications)
ClassificationSegmentationComputational EfficiencyRepresentation LearningPoint Cloud
🎯 What it does: Propose CloudMamba, a Mamba-based point cloud analysis network that integrates sequence expansion and merging, chained bidirectional Mamba, and grouped selective state space model (GS6), achieving efficient long-range modeling.
CLUENet: Cluster Attention Makes Neural Networks Have Eyes
Xiangshuai Song (National University of Defense Technology), Chang Tang (Huazhong University of Science and Technology)
ClassificationExplainability and InterpretabilityConvolutional Neural NetworkImageBenchmark
🎯 What it does: Proposed CLUENet clustering attention network to address the shortcomings of traditional convolutional/attention models in modeling irregular spatial patterns and interpretability.
CLUHCS:Dual-View Contrastive Learning Enabled Unsupervised Heterogeneous Community Search with Meta-Path Behavior Modeling
Xiaoqin Xie (Harbin Engineering University), Wu Yang (Harbin Engineering University)
Representation LearningGraph Neural NetworkTransformerContrastive LearningGraph
🎯 What it does: Propose an unsupervised dual-view contrastive learning framework named CLUHCS for community search in heterogeneous networks.
Clustering with Self-Learned Graph Regression
Lai Wei (Shanghai Maritime University), Jin Liu (Shanghai Maritime University)
OptimizationRepresentation LearningGraph Neural NetworkImage
🎯 What it does: A self-learning graph regression (SGR) method is proposed, which uses the affinity graph itself to construct regularization for graph clustering, significantly improving clustering quality.
CMedBench: A Comprehensive Benchmark for Efficient Medical Large Language Models
Shengbo Gao, Xianglong Liu (NingboTech University)
Computational EfficiencyTransformerLarge Language ModelTextBiomedical DataBenchmark
🎯 What it does: Proposed the CMedBench benchmark, systematically evaluating the performance and efficiency of compressed large language models (LLMs) in medical scenarios, covering five dimensions: knowledge, application, credibility, cross-combination, and computational efficiency.
CMID: Towards Medical Visual Question Answering via Contrastive Mutual Information Decoding
Zhihong Zhu (Tencent Jarvis Lab), Xian Wu (Tencent Jarvis Lab)
Vision Language ModelContrastive LearningMultimodalityBiomedical Data
🎯 What it does: During the inference phase, contrastive mutual information decoding (CMID) is applied to the attention of large vision-language models (LVLM) to improve the diagnostic accuracy of medical visual question answering (Med-VQA).
CMMCoT: Enhancing Complex Multi-Image Comprehension via Multi-Modal Chain-of-Thought and Memory Augmentation
Guanghao Zhang (Alibaba Group), Hao Jiang (Alibaba Group)
TransformerVision Language ModelImageMultimodalityBenchmarkRetrieval-Augmented GenerationChain-of-Thought
🎯 What it does: Propose the CMMCoT framework, achieving slow thinking understanding of multi-image data through multi-step visual-textual interleaved chain reasoning.
CNM-UNet: Continuous Ordinary Differential Equations for Medical Image Segmentation
Tianqi Xu (Sichuan University), Tao He (Sichuan University)
SegmentationConvolutional Neural NetworkPrompt EngineeringImageBiomedical DataOrdinary Differential Equation
🎯 What it does: Proposed a UNet with Continuous Neural Memory ODE (CNM-UNet), achieving efficient and lightweight medical image segmentation by replacing the entire original UNet decoder with a single CNM-Block;
CO-Bench: Benchmarking Language Model Agents in Algorithm Search for Combinatorial Optimization
Weiwei Sun (Carnegie Mellon University), Yiming Yang (Carnegie Mellon University)
OptimizationLarge Language ModelAgentic AIBenchmarkChain-of-Thought
🎯 What it does: This study proposes the CO-Bench benchmark, which includes 36 real-world combinatorial optimization (CO) problems, and systematically evaluates them on 15 LLMs and 9 agent frameworks, comparing their performance with human-designed classical solvers;
Co-EPG: A Framework for Co-Evolution of Planning and Grounding in Autonomous GUI Agents
Yuan Zhao (Alibaba Cloud Computing), Hao Henry Wang (Alibaba Cloud Computing)
Robotic IntelligenceLarge Language ModelReinforcement LearningMixture of ExpertsVision Language ModelMultimodality
🎯 What it does: Proposed the Co-EPG framework to achieve self-iterative co-evolution between planning and localization models, forming a positive feedback loop training process.
Co-Layout: LLM-driven Co-optimization for Interior Layout
Chucheng Xiang (University of Science and Technology of China), Ligang Liu (University of Science and Technology of China)
OptimizationTransformerLarge Language ModelTextMesh
🎯 What it does: This paper proposes an end-to-end framework for indoor design, leveraging large language models (LLMs) to extract spatial constraints from textual requirements, and then simultaneously optimizing room layouts and furniture arrangements through grid-based integer programming.
CO²IF: Language-Bridging Hyperspectral-Multispectral Image Fusion with Coordinated and Cross-modal Optimal Transport
Mingjin Zhang (Xidian University), Fei Gao (Xidian University)
RestorationSuper ResolutionVision Language ModelContrastive LearningImageMultimodality
🎯 What it does: Designed and implemented a language-guided high-resolution spectral image fusion framework, CO IF, capable of reconstructing high-quality HR-HSI by integrating low-resolution hyperspectral images with high-resolution multispectral images.
Coarse-to-Fine Open-Set Graph Node Classification with Large Language Models
Xueqi Ma (University of Melbourne), James Bailey (University of Melbourne)
ClassificationAnomaly DetectionGraph Neural NetworkTransformerLarge Language ModelPrompt EngineeringTextGraph
🎯 What it does: Proposed a coarse-to-fine hierarchical open-set graph node classification framework (CFC) based on large language models (LLMs), achieving simultaneous detection and classification of out-of-distribution (OOD) nodes.
CoCo-MILP: Inter-Variable Contrastive and Intra-Constraint Competitive MILP Solution Prediction
Tianle Pu (National University of Defense Technology), Changjun Fan (National University of Defense Technology)
OptimizationGraph Neural NetworkContrastive LearningGraphBenchmark
🎯 What it does: A CoCo-MILP framework is proposed for mixed integer linear programming (MILP) problems, which predicts high-quality initial solutions by contrastive learning on variables and constructing competitive mechanisms within constraints, thereby accelerating traditional solvers.
CoCoLIT: ControlNet-Conditioned Latent Image Translation for MRI to Amyloid PET Synthesis
Alec Sargood (University College London), Daniel C. Alexander (University College London)
Image TranslationData SynthesisDiffusion modelAuto EncoderImageBiomedical DataMagnetic Resonance ImagingPositron Emission TomographyAlzheimer's Disease
🎯 What it does: Achieving synthetic translation from structural MRI to Alzheimer's Aβ PET scans via the ControlNet-conditional latent diffusion model CoCoLIT.
Codebook-Centric Deep Hashing: End-to-End Joint Learning of Semantic Hash Centers and Neural Hash Function
Shuo Yin (Beijing University of Posts and Telecommunications), Dell Zhang (China Telecom)
RetrievalConvolutional Neural NetworkImageBenchmark
🎯 What it does: Propose an end-to-end deep hashing framework CRH that dynamically reallocates hash centers during training, achieving joint optimization of hash centers and hash functions.
Codebook-Empowered Analysis-Friendly Extreme Underwater Image Compression
JianHao Wu, Qiuping Jiang (City University of Hong Kong)
CompressionAuto EncoderContrastive LearningImageBenchmark
🎯 What it does: Propose a full-process machine vision-friendly underwater image compression framework based on vector quantization (VQ) codebooks, covering three stages: feature extraction, compression, and reconstruction, while maintaining excellent machine vision performance at low bitrates.
CoEvo: Continual Evolution of Symbolic Solutions Using Large Language Models
Ping Guo (City University of Hong Kong), Xi Lin (City University of Hong Kong)
OptimizationTransformerLarge Language ModelBenchmarkPhysics Related
🎯 What it does: Propose the CoEvo framework by combining large language models with evolutionary search, achieving continuous evolution and knowledge management of symbolic solutions.
CoEvoer: Collaborative Evolution Transformer for Upper-Body Expressive Human Pose and Shape Estimation
Yuxiang Zhao (Sun Yat-sen University), Huan Zhao (Sun Yat-sen University)
Pose EstimationTransformerImage
🎯 What it does: Propose a one-stage collaborative evolution transformer, CoEvoer, specifically designed for upper-body expression pose and shape estimation, significantly enhancing mutual compensation and correction between facial, hand, and torso regions.
CoFact: Dynamic Coordination of Attention Heads for Improving Factual Consistency in LLMs
Shike Li (Shanxi University), Hu Zhang (Shanxi University)
TransformerLarge Language ModelText
🎯 What it does: CoFact enhances factual consistency in large language models by dynamically coordinating the behavior of multi-head attention heads during inference.
Cog-RAG: Cognitive-Inspired Dual-Hypergraph with Theme Alignment Retrieval-Augmented Generation
Hao Hu (Xi'an Jiaotong University), Shaoyi Du (Xi'an Jiaotong University)
RetrievalGraph Neural NetworkLarge Language ModelTextBiomedical DataAgriculture RelatedRetrieval-Augmented Generation
🎯 What it does: By employing a dual hypergraph structure and a two-stage retrieval strategy, enhancing the generation quality of RAG models in knowledge-intensive tasks
CoGenSAM: Codebook-Interactive Generative Labeling for Adapting SAM to Crack Segmentation
Zhuangzhuang Chen (Shenzhen University), Jianqiang Li (Shenzhen University)
RestorationSegmentationAuto EncoderContrastive LearningImage
🎯 What it does: This paper proposes the CoGenSAM framework, achieving SAM refinement for crack segmentation without manual annotation. It generates high-quality crack masks automatically using generated labels and a recovery model, eliminating the cost of manual annotation.
Cognitive Enhancement Chain-of-Thought Towards Enhancing Style Learning and Content Preservation for Long Style Transfer
Lianwei Wu (Northwestern Polytechnical University), Xianghua Li (Northwest Polytechnical University)
GenerationTransformerLarge Language ModelPrompt EngineeringTextChain-of-Thought
🎯 What it does: This paper proposes a long-text style transfer framework based on cognitive-enhanced chain-of-thought (CeCoT), which leverages LLM to progressively rewrite the source text while preserving semantics, and learns the target style through cognitive chain-of-thought to achieve long-text style transfer.
CogniTrust: Cognitive Memory-Driven Verifiable Supervision for Robust Hashing
Yiyang Gu (Peking University), Ming Zhang (Peking University)
RetrievalConvolutional Neural NetworkDiffusion modelAuto EncoderImage
🎯 What it does: Propose the CogniTrust framework, which utilizes three memory mechanisms (episodic memory, semantic memory, and reconstructive memory) to correct supervision and enhance retrieval performance in multi-label hashing tasks with label noise.
CoGrad3D: Spatially-Coupled Timestep Optimization with Orthogonal Gradient Fusion for 3D Generation
Haoyang Tong (Chinese Academy of Sciences), Ran He (Chinese Academy of Sciences)
GenerationData SynthesisDiffusion modelScore-based ModelTextMesh
🎯 What it does: This paper proposes a unified text-to-3D generation framework called CoGrad3D, which significantly improves the performance of generative models in terms of detail quality and multi-view consistency by combining adaptive region sampling, hybrid time step scheduling, and gradient fusion techniques on the basis of Score Distillation Sampling (SDS).
COGS: A Causal Representation Learning Framework for Out-of-Distribution Generalization in Time Series
Xinxin Song (Tsinghua University), Jinli Suo (Tsinghua University)
Domain AdaptationRepresentation LearningAuto EncoderContrastive LearningTime Series
🎯 What it does: Propose the CO-GS framework, which leverages causal representation learning and unsupervised domain discovery to achieve out-of-distribution (OOD) generalization for time series
CogStream: Context-guided Streaming Video Question Answering
Zicheng Zhao (Shanghai Jiao Tong University), Huabin Liu (Shanghai Jiao Tong University)
RetrievalCompressionTransformerLarge Language ModelVision Language ModelVideoTextMultimodalityRetrieval-Augmented Generation
🎯 What it does: Proposes the CogStream task and a corresponding semi-automatically created dataset, challenging models to identify and utilize the most relevant historical context in real-time videos for question answering.
CoherenDream: Boosting Holistic Text Coherence in 3D Generation via Multimodal Large Language Models Feedback
Chenhan Jiang (Hong Kong University of Science and Technology), Dit-Yan Yeung (Shanghai Jiao Tong University)
GenerationData SynthesisTransformerLarge Language ModelVision Language ModelScore-based ModelNeural Radiance FieldImageTextMultimodality
🎯 What it does: Propose the CoherenDream framework, which utilizes Score Distillation Sampling and feedback from multi-modal large language models to generate 3D content consistent with text.
COIN: Uncertainty-Guarding Selective Question Answering for Foundation Models with Provable Risk Guarantees
Zhiyuan Wang (University of Electronic Science and Technology of China), Kaidi Xu (Drexel University)
TransformerLarge Language ModelVision Language ModelTextMultimodality
🎯 What it does: Proposes the COIN framework, which sets thresholds during answer generation by the base model and filters high-uncertainty samples to ensure the error rate does not exceed the user-specified FDR.
Collaborative Dual Representations for Semi-Supervised Partial Label Learning
Wei-Xuan Bao (Southeast University), Min-Ling Zhang (Lenovo Group Ltd)
ClassificationRepresentation LearningConvolutional Neural NetworkImage
🎯 What it does: Proposed a semi-supervised partial label learning framework named CODUAL, which jointly learns the predictive class distribution and low-dimensional embeddings for each sample, and enhances model performance through cooperative updates between the two.
Collaborative Enhancement of Large and Small Models for Question Answering via Dual Knowledge Transfer
Shaofei Wang (Capital Normal University), Wenlong Chen (Capital Normal University)
Knowledge DistillationTransformerLarge Language ModelPrompt EngineeringText
🎯 What it does: Achieve collaborative enhancement between large and small models through a bidirectional knowledge transfer (DKT) method, specifically including two steps: small4large (confidence-based prompting to enhance the large model) and large4small (reflective knowledge distillation to improve the small model), with iterative performance improvement between the two steps.
Collaborative Feature Matching with Progressive Correspondence Learning
Xin Liu (Nankai University), Jufeng Yang (Nankai University)
Pose EstimationGraph Neural NetworkTransformerImage
🎯 What it does: Proposes an end-to-end collaborative feature matching framework CFM, comprising a keypoint learning module and a correspondence learning module, which progressively enhance matching quality through mutual reinforcement.
Collaborative LLM Numerical Reasoning with Local Data Protection
Min Zhang (Virginia Tech), Haozhu Wang (AWS AI)
Federated LearningSafty and PrivacyComputational EfficiencyKnowledge DistillationLarge Language ModelPrompt EngineeringTextMultimodalityFinance RelatedChain-of-Thought
🎯 What it does: Propose a collaborative reasoning framework that sends queries for numerical reasoning tasks to remote models after anonymizing them through topic transformation and numerical masking, and then reconstructs answers locally using reusable Python tools.
Collaborative Representation Learning for Alignment of Tactile, Language, and Vision Modalities
Yiyun Zhou (Zhejiang University), Jingyuan Chen (National University of Defense Technology)
Representation LearningTransformerContrastive LearningMultimodality
🎯 What it does: This paper proposes a multimodal collaborative representation learning framework called TLV-CoRe based on CLIP, which can map tactile, linguistic, and visual modalities into a unified latent space to achieve cross-modal alignment.
Collaborative Transformers with Multi-Level Forensic Attention for Image Manipulation Localization
Jiwei Zhang (Beijing University Of Posts And Telecommunications), Shaozhang Niu (China Mobile Internet Co)
Anomaly DetectionTransformerImage
🎯 What it does: Proposed a collaborative Transformer framework called Co-Transformers for precisely locating image tampering regions.
Collaboratively “Copy & Paste” 2D-3D Features for Complex Video-to-Video Motion Editing
Jia-Xing Zhong (ByteDance Inc), Li Zhang (ByteDance Inc)
Image TranslationGenerationPose EstimationKnowledge DistillationConvolutional Neural NetworkDiffusion modelVideo
🎯 What it does: This paper proposes a collaborative 'copy-and-paste' 2D-3D feature framework for complex video-to-video human motion editing;