AAAI 2025 Papers — Page 19
AAAI Conference on Artificial Intelligence · 3028 papers
Mitigating Social Bias in Large Language Models: A Multi-Objective Approach Within a Multi-Agent Framework
Zhenjie Xu (Sun Yat-sen University), Zhichao Lu (City University of Hong Kong)
OptimizationTransformerLarge Language ModelPrompt EngineeringText
🎯 What it does: A multi-objective, multi-agent framework (MOMA) is proposed to mitigate social biases in large language models, primarily through causal interventions on the representations of social groups in the input using two agents: masking and balancing.
Mixed-Curvature Multi-Modal Knowledge Graph Completion
Yuxiao Gao (Beihang University), Fuzhen Zhuang (Beihang University)
OptimizationKnowledge DistillationRepresentation LearningGraph Neural NetworkMultimodalityGraph
🎯 What it does: This paper proposes the MCKGC method, which utilizes three types of curvature spaces—hyperplane, spherical, and Euclidean space—to embed the structure, image, and text information of multimodal knowledge graphs, and achieves cross-modal and cross-space information integration through a progressive fusion mechanism.
Mixture of Experts as Representation Learner for Deep Multi-View Clustering
Yunhe Zhang (University of Macau), See-Kiong Ng (National University of Singapore)
Representation LearningMixture of ExpertsAuto EncoderImage
🎯 What it does: A deep multi-view clustering framework DMVC-CE based on Mixture of Experts (MoE) is proposed, which utilizes a gating network to dynamically select experts for learning multi-view representations.
Mixture of Experts Based Multi-Task Supervise Learning from Crowds
Tao Han (Zhejiang Gongshang University), Yili Fang (Zhejiang Gongshang University)
Mixture of ExpertsImageText
🎯 What it does: A new paradigm of multi-task supervised learning - crowds, called MLC, is proposed, and a Mixture-of-Experts model MMLC is constructed on it. Two aggregation strategies are further provided: utilizing oracle worker discovery (MMLC-owf) and data filling based on (MMLC-df) to achieve high-quality true labeling.
Mixture of Knowledge Minigraph Agents for Literature Review Generation
Zhi Zhang (Hong Kong Polytechnic University), Jiannong Cao (Education University of Hong Kong)
GenerationTransformerLarge Language ModelPrompt EngineeringMixture of ExpertsTextReview/Survey PaperBenchmark
🎯 What it does: A framework for Collaborative Knowledge Minimal Graph Agents (CKMAs) is proposed for the automated generation of academic literature reviews, which includes a Knowledge Minimal Graph Construction Agent (KMCA) and a Multi-Path Summary Agent (MPSA).
Mixture of Online and Offline Experts for Non-Stationary Time Series
Zhilin Zhao (Sun Yat-sen University), Yuanyu Wan (Zhejiang University)
Mixture of ExpertsTime Series
🎯 What it does: A hybrid expert framework MOOE is proposed, which combines offline experts and online experts to address the time series forecasting problem with multiple segments of non-stationary distributions.
Mixture-of-Attack-Experts with Class Regularization for Unified Physical-Digital Face Attack Detection
Shunxin Chen (Nanjing University of Posts and Telecommunications), Zhen Lei (Computer Vision Center)
ClassificationRecognitionAnomaly DetectionTransformerMixture of ExpertsVision Language ModelContrastive LearningImage
🎯 What it does: The MoAE-CR framework is proposed, incorporating the Mixture-of-Attack-Experts and Class Regularization modules into the CLIP visual-language model to achieve unified detection of digital and physical facial attacks.
Mjölnir: Breaking the Shield of Perturbation-Protected Gradients via Adaptive Diffusion
Xuan Liu (Hong Kong Polytechnic University), Kaiwei Lin (Hong Kong Polytechnic University)
Federated LearningSafty and PrivacyDiffusion modelImage
🎯 What it does: In federated learning, a diffusion model is used to denoise gradients disturbed by differential privacy noise, thereby recovering and leaking the private data of clients.
ML-GOOD: Towards Multi-Label Graph Out-Of-Distribution Detection
Tingyi Cai (Zhejiang Normal University), Qionghao Huang (Zhejiang Normal University)
Anomaly DetectionGraph Neural NetworkGraph
🎯 What it does: Research on the OOD detection problem in multi-label graph data, proposing the ML-GOOD framework;
MLAAN: Scaling Supervised Local Learning with Multilaminar Leap Augmented Auxiliary Network
Yuming Zhang (Southeast University), Changpeng Cai (Southeast University)
ClassificationConvolutional Neural NetworkSupervised Fine-TuningImage
🎯 What it does: A Multi-Layer Jump Augmented Auxiliary Network (MLAAN) is proposed to address the issues of local learning lacking global information and having a narrow perspective;
MLC-NC: Long-Tailed Multi-Label Image Classification Through the Lens of Neural Collapse
Zijian Tao (Nanjing University of Aeronautics and Astronautics), Songcan Chen (Nanjing University of Aeronautics and Astronautics)
ClassificationConvolutional Neural NetworkContrastive LearningImage
🎯 What it does: A neural folding (NC) driven method for long-tail multi-label image classification, called MLC-NC, is proposed. It utilizes fixed equiangular tight frame (ETF) label embedding to guide class feature learning, reducing intra-class variance and using a binarized fixed ETF classifier to suppress classification bias.
MM-CamObj: A Comprehensive Multimodal Dataset for Camouflaged Object Scenarios
Jiacheng Ruan (Shanghai Jiao Tong University), Yuzhuo Fu (Institute for Advanced Algorithms Research)
Object DetectionSegmentationTransformerLarge Language ModelSupervised Fine-TuningVision Language ModelImageMultimodalityBenchmark
🎯 What it does: Constructed the MM-CamObj two subsets (CamObj-Align and CamObj-Instruct) and trained a specialized CamObj-Llava based on this, proposing CamObj-Bench for evaluating multimodal models in camouflage scenarios.
MM-Mixing: Multi-Modal Mixing Alignment for 3D Understanding
Jiaze Wang (Central South University), Pheng-Ann Heng (Chinese University of Hong Kong)
RecognitionRetrievalTransformerContrastive LearningMultimodalityPoint Cloud
🎯 What it does: Proposes the MM-Mixing multimodal mixing alignment framework, which enhances 3D understanding performance through a two-stage feature-level and input-level mixing.
MM-Tracker: Motion Mamba for UAV-platform Multiple Object Tracking
Mufeng Yao (Fudan University), Jon Atli Benediktsson (University of Iceland)
Object DetectionObject TrackingOptical FlowVideo
🎯 What it does: Achieving multi-object tracking in drone videos, proposing two key components: the Motion Mamba module and Motion Margin Loss.
mmFAS: Multimodal Face Anti-Spoofing Using Multi-Level Alignment and Switch-Attention Fusion
Geng Chen (Beijing University of Posts and Telecommunications), Miaohui Wang (Nanjing University of Posts and Telecommunications)
ClassificationRecognitionAnomaly DetectionTransformerContrastive LearningImageMultimodality
🎯 What it does: A multi-modal face anti-spoofing framework mmFAS is proposed, which combines multi-level feature alignment and switch-attention fusion to achieve joint discrimination of RGB, depth, and infrared modalities.
MMGDreamer: Mixed-Modality Graph for Geometry-Controllable 3D Indoor Scene Generation
Zhifei Yang (Peking University), Guangyao Zhai (Technical University of Munich)
GenerationData SynthesisGraph Neural NetworkDiffusion modelMultimodalityGraph
🎯 What it does: Generate 3D indoor scenes with precisely controllable geometric shapes through a mixed-modal graph (MMG) and a dual-branch diffusion model;
MMPF: Multi-Modal Perception Framework for Abnormal Medical Condition Detection
Chuyi Zhong (Fudan University), Lihua Zhang (Fudan University)
Anomaly DetectionConvolutional Neural NetworkImageMultimodality
🎯 What it does: This paper proposes a multi-modal perception framework (MMPF) for non-invasive detection of abnormal medical conditions from images.
MOCID: Motion Context and Displacement Information Learning for Moving Infrared Small Target Detection
Mingjin Zhang (Xidian University), Jing Zhang (Wuhan University)
Object DetectionOptical FlowVideo
🎯 What it does: Proposes the MOCID system, which improves the detection of small moving infrared targets by bidirectionally fusing clip-level motion context with frame-level displacement information.
Modality-Aware Shot Relating and Comparing for Video Scene Detection
Jiawei Tan (Chongqing University), Zhilong Ou (Chongqing University)
Object DetectionSegmentationConvolutional Neural NetworkGraph Neural NetworkVideoMultimodality
🎯 What it does: A MASRC framework based on the multimodal (entity and scene) interrelationship is proposed, which captures long-term and short-term associations using graph networks and detects video scene boundaries through multi-shot comparison.
Modality-Independent Graph Neural Networks with Global Transformers for Multimodal Recommendation
Jun Hu (National University of Singapore), Yinwei Wei (Shandong University)
Recommendation SystemGraph Neural NetworkTransformerContrastive LearningTextMultimodality
🎯 What it does: A multimodal recommendation framework based on modality-independent receptive fields in graph neural networks and a sampling-based global Transformer is proposed, which can simultaneously utilize text, visual, and learnable embedding information to enhance the expressiveness of user-item representations.
Model Lineage Closeness Analysis
Chen Tang (University of Science and Technology of China), Xiang-Yang Li
ClassificationOptimizationKnowledge DistillationGenerative Adversarial NetworkImageBenchmark
🎯 What it does: This paper proposes a method for measuring model change lineage, which can quantify the degree of modification between two models (Lineage closeness).
ModelDiff: Symbolic Dynamic Programming for Model-Aware Policy Transfer in Deep Q-Learning
Xiaotian Liu (University of Toronto), Scott Sanner (University of Toronto)
Domain AdaptationReinforcement Learning
🎯 What it does: A model difference-based symbolic dynamic programming method called ModelDiff is proposed to achieve model-aware policy transfer in deep Q-learning.
Modeling All Response Surfaces in One for Conditional Search Spaces
Jiaxing Li (China University of Petroleum), Dacheng Tao (Nanyang Technological University)
OptimizationHyperparameter SearchNeural Architecture SearchTransformerReinforcement LearningTabular
🎯 What it does: A unified attention mechanism Bayesian optimization framework, AttnBO, is proposed, which constructs a unified Gaussian process model using structure-aware embeddings and a Transformer encoder to address the hyperparameter optimization problem in conditional search spaces.
Modeling Inter-Intra Heterogeneity for Graph Federated Learning
Wentao Yu (Nanjing University of Science and Technology), Chen Gong (Shanghai Jiao Tong University)
Federated LearningGraph Neural NetworkAuto EncoderGraph
🎯 What it does: A new graph federated learning method called FedIIH is proposed, which can simultaneously consider the inter-heterogeneity of graph data across subgraphs and the intra-heterogeneity within subgraphs, enhancing the performance of distributed graph models.
Modeling Latent Non-Linear Dynamical System over Time Series
Ren Fujiwara (Osaka University), Yasushi Sakurai (Osaka University)
Time Series
🎯 What it does: A method called LaNoLem is proposed for jointly estimating latent states and nonlinear dynamic equations given time series data.
MoDiTalker: Motion-Disentangled Diffusion Model for High-Fidelity Talking Head Generation
Seyeon Kim (Korea University), Seungryong Kim (KAIST)
GenerationData SynthesisTransformerDiffusion modelVideoAudio
🎯 What it does: Proposes MoDiTalker, a two-stage speech-driven lip-sync generation framework based on diffusion models.
Modular-Cam: Modular Dynamic Camera-view Video Generation with LLM
Zirui Pan (Tsinghua University), Wenwu Zhu (Tsinghua University)
GenerationData SynthesisTransformerLarge Language ModelDiffusion modelVideoText
🎯 What it does: The Modular-Cam framework is proposed, achieving text-based dynamic camera perspective video generation for multiple scenes.
MoE-LPR: Multilingual Extension of Large Language Models Through Mixture-of-Experts with Language Priors Routing
Hao Zhou (Nanjing University), Jiajun Chen (Alibaba International Digital Commerce)
Large Language ModelMixture of ExpertsText
🎯 What it does: This paper proposes a two-stage multilingual expansion method called MoE-LPR, which upgrades the pre-trained LLM to a Mixture-of-Experts architecture. It first trains new experts on the newly added languages and then trains a router using a small amount of original language replay data with a Language Priority Routing (LPR) mechanism, enhancing the capabilities of the new languages while maintaining the performance of the original languages.
MOL-Mamba: Enhancing Molecular Representation with Structural & Electronic Insights
Jingjing Hu (Hefei University of Technology), Meng Wang (Hefei University of Technology)
Representation LearningDrug DiscoveryGraph Neural NetworkTransformerGraph
🎯 What it does: This paper proposes the MOL-Mamba framework, which enhances molecular representation by integrating structural and electronic information on molecular graphs.
MoLE:Decoding by Mixture of Layer Experts Alleviates Hallucination in Large Vision-Language Models
Tian Liang (Zhejiang University), Qiang Zhu (Zhejiang University)
GenerationTransformerLarge Language ModelMixture of ExpertsVision Language ModelImageText
🎯 What it does: A training-free decoding method called MoLE is proposed, which significantly reduces the phenomenon of hallucination in text generation by selecting experts from different layers of a large visual language model and generating collaboratively.
Momentum Pseudo-Labeling for Weakly Supervised Phrase Grounding
Dongdong Kuang (Beihang University), Jaein Kim (Beihang University)
RecognitionObject DetectionContrastive LearningImageText
🎯 What it does: Proposes the Momentum Pseudo Labeling (MPL) framework, which utilizes a momentum model to achieve global pseudo label updates and considers false negative samples in non-matching image-text pairs during contrastive learning, thereby enhancing weakly supervised phrase localization performance.
Monitoring Primitive Interactions During the Training of DNNs
Jie Ren (Shanghai Jiao Tong University), Quanshi Zhang (Shanghai Jiao Tong University)
Explainability and InterpretabilityComputational EfficiencyConvolutional Neural NetworkRecurrent Neural NetworkSupervised Fine-TuningImageTextPoint CloudTabular
🎯 What it does: This paper constructs a framework that can adequately explain the learning process of deep networks with a minimal number of interactions by redefining interaction relationships on the principal components of intermediate layer features and tracking the dynamics of these interactions during training.
MonoBox: Tightness-Free Box-Supervised Polyp Segmentation Using Monotonicity Constraint
Qiang Hu (Wuhan National Laboratory for Optoelectronics), Zhiwei Wang (Wuhan National Laboratory for Optoelectronics)
Object DetectionSegmentationImage
🎯 What it does: A box-supervised polyp segmentation method named MonoBox is proposed, which can achieve pixel-level segmentation without satisfying the box compactness assumption.
Moonshine: Distilling Game Content Generators into Steerable Generative Models
Yuhe Nie (New York University), Julian Togelius (New York University)
GenerationKnowledge DistillationLarge Language ModelDiffusion modelText
🎯 What it does: A controllable text-conditioned game map generation model was created by combining the generation results of traditional procedural content generation (PCG) algorithms with text descriptions automatically annotated by LLM.
More Text, Less Point: Towards 3D Data-Efficient Point-Language Understanding
Yuan Tang (Huazhong University of Science and Technology), Min Chen (South China University of Technology)
ClassificationGenerationTransformerLarge Language ModelSupervised Fine-TuningPrompt EngineeringTextMultimodalityPoint Cloud
🎯 What it does: The GreenPLM model is proposed, achieving language understanding and generation for 3D objects with only 12% of 3D training samples or even without 3D data, through a three-stage training process, 0M-Pooling, and large-scale text data.
MoRe: Class Patch Attention Needs Regularization for Weakly Supervised Semantic Segmentation
Zhiwei Yang (Fudan University), Zhijian Song (Fudan University)
SegmentationTransformerContrastive LearningImage
🎯 What it does: Proposes the MoRe framework, which utilizes class-patch attention from Vision Transformer and generates more refined pseudo-labels for weakly supervised semantic segmentation through graph regularization and CAM-guided attention regularization.
MORE: Molecule Pretraining with Multi-Level Pretext Task
Yeongyeong Son (Pusan National University), Sunyoung Kwon (Pusan National University)
Representation LearningDrug DiscoveryGraph Neural NetworkGraphBiomedical Data
🎯 What it does: A multi-level molecular graph pre-training framework called MORE is proposed, which combines self-supervised tasks of nodes, subgraphs, graphs, and 3D perspectives;
MOS: Model Surgery for Pre-Trained Model-Based Class-Incremental Learning
Hai-Long Sun (Nanjing University), Han-Jia Ye (Nanjing University)
ClassificationRecognitionTransformerImageBenchmark
🎯 What it does: The MOS (Model Surgery) method is proposed, utilizing task-specific adapters, adapter fusion, and a self-correcting retrieval mechanism to address the catastrophic forgetting problem encountered by pre-trained models in class-incremental learning.
Motif Guided Graph Transformers with Combinatorial Skeleton Prototype Learning for Skeleton-Based Person Re-Identification
Haocong Rao (Nanyang Technological University), Chunyan Miao (Nanyang Technological University)
RecognitionRetrievalGraph Neural NetworkTransformerContrastive LearningVideoGraph
🎯 What it does: This paper proposes a pedestrian re-identification method based on 3D skeletons called MoCos, which integrates graphical synapses and combinatorial skeleton prototype learning to enhance skeleton representation.
Motif-aware Graph Neural Networks for Networked Time Series Imputation
Nourhan Ahmed (University of Hildesheim), Lars Schmidt-Thieme (University of Hildesheim)
Graph Neural NetworkAuto EncoderGraphTime Series
🎯 What it does: This study proposes Motif-GNN, a method for missing value imputation that utilizes graph neural networks combined with high-order network substructures (network motifs) in time series.
Motif-Oriented Representation Learning with Topology Refinement for Drug-Drug Interaction Prediction
Ran Zhang (Computer Network Information Center, Chinese Academy of Sciences), Pengfei Wang (Computer Network Information Center, Chinese Academy of Sciences)
Representation LearningDrug DiscoveryGraph Neural NetworkContrastive LearningGraphBiomedical Data
🎯 What it does: Predicting interactions by extracting motif information within molecules and combining it with DDI network topology.
Motion Artifact Removal in Pixel-Frequency Domain via Alternate Masks and Diffusion Model
Jiahua Xu (Xidian University), Xinbo Gao (Chongqing University of Posts and Telecommunications)
RestorationDiffusion modelImageBiomedical DataMagnetic Resonance Imaging
🎯 What it does: This paper proposes an unsupervised motion artifact removal method called PFAD, which utilizes pixel-frequency alternating masks to guide a diffusion model in recovering clean MRI images.
Motion Decoupled 3D Gaussian Splatting for Dynamic Object Representation
Xiao Hu (University of Ottawa), Jochen Lang (University of Ottawa)
RestorationGenerationGaussian SplattingPoint Cloud
🎯 What it does: A dynamic 3D Gaussian rendering framework M5D-GS is proposed, which decouples motion and deformation, achieving high-quality reconstruction and real-time rendering of significantly moving objects using a monocular camera.
Motion Prior Knowledge Learning with Homogeneous Language Descriptions for Moving Infrared Small Target Detection
Shengjia Chen (University of Electronic Science and Technology of China), Mao Ye (University of Electronic Science and Technology of China)
Object DetectionConvolutional Neural NetworkRecurrent Neural NetworkGraph Neural NetworkVision Language ModelVideoTextMultimodality
🎯 What it does: This paper proposes a motion prior knowledge learning framework MoPKL based on audiovisual fusion, which guides the visual channel to learn the motion characteristics of small infrared targets through homogeneous language descriptions, thereby achieving fine detection of moving small targets.
Motion-adaptive Transformer for Event-based Image Deblurring
Senyan Xu (University of Science and Technology of China), Yan Chen (University of Science and Technology of China)
RestorationTransformerImage
🎯 What it does: A Motion-Adaptive Transformer (MAT) is proposed, which improves the fusion and attention mechanism of events and images by utilizing the spatial motion information provided by event cameras, achieving high-quality motion blur removal.
Motion-aware Contrastive Learning for Temporal Panoptic Scene Graph Generation
Thong Thanh Nguyen (Nanyang Technological University), Anh Tuan Luu (Nanyang Technological University)
Object DetectionSegmentationGenerationTransformerContrastive LearningVideoPoint Cloud
🎯 What it does: This paper studies the generation of temporal panoptic scene graphs and proposes a motion-aware contrastive learning framework to improve the representation of entity mask tubes.
Motion-Zero: A Zero-Shot Trajectory Control Framework of Moving Object for Diffusion-Based Video Generation
Changgu Chen (East China Normal University), Yang Li (East China Normal University)
GenerationData SynthesisDiffusion modelVideoText
🎯 What it does: A zero-shot framework called Motion-Zero is proposed to achieve precise control over the trajectories of single objects in video diffusion models.
MotionCraft: Crafting Whole-Body Motion with Plug-and-Play Multimodal Controls
Yuxuan Bian (Chinese University of Hong Kong), Qiang Xu (Chinese University of Hong Kong)
GenerationData SynthesisPose EstimationTransformerDiffusion modelMultimodality
🎯 What it does: This paper presents MotionCraft, a unified diffusion Transformer that can generate complete body movements through multimodal control using text, speech, or music.
Move and Act: Enhanced Object Manipulation and Background Integrity for Image Editing
Pengfei Jiang (Xiamen University), Fei Chao (Xiamen University)
Image TranslationObject DetectionGenerationDiffusion modelImage
🎯 What it does: This paper proposes a fine-tuning-free image consistency editing method called Move&Act, which can simultaneously control the actions and generated positions of the edited objects while maintaining the integrity of the background during the editing process.
MP: Endowing Large Language Models with Lateral Thinking
Tian Bai (Jilin University), Haitao Yu (University of Tsukuba)
TransformerLarge Language ModelPrompt EngineeringTextChain-of-Thought
🎯 What it does: A metacognitive prompting (MP) method is proposed to enable large language models to possess lateral thinking abilities;
MPQ-DM: Mixed Precision Quantization for Extremely Low Bit Diffusion Models
Weilun Feng (Institute of Computing Technology Chinese Academy of Sciences), Michele Magno (ETH Zurich)
GenerationData SynthesisKnowledge DistillationDiffusion modelImage
🎯 What it does: A method called MPQ-DM is proposed, which implements mixed precision quantization and time-smooth distillation for extremely low-bit quantization of diffusion models.
MPTSNet: Integrating Multiscale Periodic Local Patterns and Global Dependencies for Multivariate Time Series Classification
Yang Mu (Technical University of Munich), Xiao Xiang Zhu (Munich Center for Machine Learning)
ClassificationConvolutional Neural NetworkTime Series
🎯 What it does: Proposed and implemented MPTSNet for multivariate time series classification, combining multi-scale periodic decomposition, CNN, and attention mechanisms to simultaneously extract local and global features.
MRBTP: Efficient Multi-Robot Behavior Tree Planning and Collaboration
Yishuai Cai (National University of Defense Technology), Ji Wang (National University of Defense Technology)
OptimizationRobotic IntelligenceLarge Language ModelTabular
🎯 What it does: A provably sound and complete behavior tree planning algorithm for multi-robot systems, MRBTP, has been designed, supporting cross-tree expansion, backup structures, intention sharing, and providing an optional LLM subtree pre-planning plugin.
mRNA2vec: mRNA Embedding with Language Model in the 5'UTR-CDS for mRNA Design
Honggen Zhang (University of Hawaii), Lipeng Lai (XtalPi Inc)
OptimizationDrug DiscoveryTransformerLarge Language ModelBiomedical Data
🎯 What it does: This paper proposes an mRNA embedding method called mRNA2vec based on data2vec, which jointly utilizes 5' UTR and CDS sequences for contextual pre-training. It enhances mRNA translation efficiency and expression level prediction through hard masking and auxiliary tasks (MFE regression, secondary structure classification).
MRR-FV: Unlocking Complex Fact Verification with Multi-Hop Retrieval and Reasoning
Liwen Zheng (Beijing University of Posts and Telecommunications), Xi Zhang (Beijing University of Posts and Telecommunications)
GenerationRetrievalGraph Neural NetworkTransformerTextRetrieval-Augmented Generation
🎯 What it does: A generative retrieval-inference framework named MRR-FV is proposed to address the multi-hop retrieval and reasoning problem in fact verification.
MSAmba: Exploring Multimodal Sentiment Analysis with State Space Models
Xilin He (Shenzhen University), Zitong Yu (University of Exeter)
ClassificationTransformerMultimodality
🎯 What it does: This paper proposes a multimodal sentiment analysis framework called MSAmba based on Mamba, which mainly implements modeling of internal sequence information of each modality and cross-modal interaction through two major modules (ISM and CHM);
MSE-Adapter: A Lightweight Plugin Endowing LLMs with the Capability to Perform Multimodal Sentiment Analysis and Emotion Recognition
Yang Yang (South China University of Technology), Yupeng Qiang (South China University of Technology)
ClassificationRecognitionTransformerLarge Language ModelTextMultimodality
🎯 What it does: A lightweight plugin MSE-Adapter is proposed, enabling large language models to perform multimodal sentiment analysis and emotion recognition tasks with approximately 2.6M–2.8M trainable parameters while maintaining their original general capabilities.
MSP-MVS: Multi-Granularity Segmentation Prior Guided Multi-View Stereo
Zhenlong Yuan (Institute of Computing Technology, Chinese Academy of Sciences), Zhaoqi Wang (Institute of Computing Technology, Chinese Academy of Sciences)
RestorationSegmentationImageBenchmark
🎯 What it does: Proposes the MSP-MVS method, which utilizes multi-granularity segmentation priors to achieve edge-constrained deformable patch matching, thereby improving the reconstruction quality of texture-sparse areas.
MSR: A Multifaceted Self-Retrieval Framework for Microscopic Cascade Prediction
Dongsheng Hong (Fuzhou University), Xiangwen Liao (Minjiang University)
RetrievalRecommendation SystemRepresentation LearningGraph Neural NetworkReinforcement LearningGraphSequential
🎯 What it does: This paper proposes a multi-perspective self-retrieval framework MSR, which constructs multi-channel GRAU encoding using social relationship graphs and behavioral preference graphs, transforming micro-level information diffusion prediction into a retrieval task, and utilizing cosine similarity for prediction.
MSSDA: Multi-Sub-Source Domain Adaptation for Diabetic Foot Neuropathy Recognition
Yan Zhong (South China University of Technology), Peiru Zhou (Jinan University)
RecognitionDomain AdaptationConvolutional Neural NetworkContrastive LearningTime SeriesBiomedical Data
🎯 What it does: This paper proposes a Multi-Source Subdomain Adaptation framework (MSSDA) that utilizes continuous plantar pressure data collected from wearable shoes to identify diabetic foot neuropathy.
MSV-PCT: Multi-Sparse-View Enhanced Transformer Framework for Salient Object Detection in Point Clouds
Zihao Wang (Beijing University of Technology), Yongjian Deng (Beijing University of Technology)
Object DetectionTransformerPoint Cloud
🎯 What it does: A multi-view enhanced Transformer framework MSV-PCT is proposed for point cloud salient object detection.
MTGA: Multi-View Temporal Granularity Aligned Aggregation for Event-Based Lip-Reading
Wenhao Zhang (Wuhan University), Jialie Shen (City University of London)
RecognitionConvolutional Neural NetworkRecurrent Neural NetworkGraph Neural NetworkVideo
🎯 What it does: This paper proposes a multi-perspective event camera lip-reading framework called MTGA, which integrates features from both event frames and time-segmented voxel graph lists, achieving high-precision recognition of lip movements through spatiotemporal granularity alignment fusion and a backend implementation using position encoding, Bi-GRU, and Self-Attention.
MTL-LoRA: Low-Rank Adaptation for Multi-Task Learning
Yaming Yang (Peking University), Yunhai Tong (Peking University)
Supervised Fine-TuningMixture of ExpertsTextMultimodality
🎯 What it does: Improved LoRA in multi-task learning to better capture task-specific and shared information;
MTVHunter: Smart Contracts Vulnerability Detection Based on Multi-Teacher Knowledge Translation
Guokai Sun (Harbin Engineering University), Liguo Zhang (Harbin Engineering University)
Anomaly DetectionKnowledge DistillationNeural Architecture SearchRecurrent Neural NetworkGraph Neural NetworkTabular
🎯 What it does: A multi-teacher bytecode vulnerability detection framework called MTVHunter is designed, combining instruction denoising and semantic completion from two major teachers to enhance the security detection of smart contract bytecode.
MUC: Mixture of Uncalibrated Cameras for Robust 3D Human Body Reconstruction
Yitao Zhu (ShanghaiTech University), Qian Wang (ShanghaiTech University)
Pose EstimationTransformerAuto EncoderImageMesh
🎯 What it does: This paper proposes a 3D human reconstruction method called MUC for uncalibrated multi-view cameras, which utilizes a single-view pre-trained encoder to generate human models and camera position information for each camera. It dynamically fuses joint and surface information from different views through a Joint Re-weighting Network (JRN) and a Surface Re-weighting Network (SRN) to obtain a detailed SMPL-X human mesh.
MUCD: Unsupervised Point Cloud Change Detection via Masked Consistency
Yue Wu (Xidian University), Qiguang Miao (Xidian University)
SegmentationAnomaly DetectionAuto EncoderContrastive LearningPoint Cloud
🎯 What it does: This paper proposes an unsupervised point cloud change detection framework called MUCD, which utilizes mask consistency learning to capture change information in multi-temporal point clouds.
Multi Agent Reinforcement Learning for Sequential Satellite Assignment Problems
Joshua Holder (University of Washington), Mehran Mesbahi (University of Washington)
OptimizationReinforcement LearningAgentic AISequential
🎯 What it does: A distributed task allocation framework called REDA based on multi-agent reinforcement learning is proposed, which addresses time-related allocation problems and achieves global optimality.
Multi-Agent Motion Planning for Differential Drive Robots Through Stationary State Search
Jingtian Yan (Carnegie Mellon University), Jiaoyang Li (Carnegie Mellon University)
OptimizationRobotic IntelligenceSimultaneous Localization and MappingBenchmark
🎯 What it does: A three-layer framework called MASS is proposed, which uses MAPF to resolve conflicts, SSIPP to search for resting states and generate feasible trajectories, and SPS to optimize speed, extending the method to lifelong multi-agent motion planning.
Multi-Apartment Rent Division
Ariel D. Procaccia (Harvard University), Shirley Zhang (Harvard University)
Optimization
🎯 What it does: This paper extends the rental allocation problem to multiple apartment scenarios, proposing and studying a new concept of fairness—Negotiated Envy-Freeness—and proving its total existence and that it can be optimized for linear objectives in polynomial time.
Multi-aspect Self-guided Deep Information Bottleneck for Multi-modal Clustering
Shizhe Hu (Zhengzhou University), Yangdong Ye (Zhengzhou University)
Representation LearningImageTextMultimodality
🎯 What it does: A multi-modal clustering method is proposed - Multi-Scale Self-Guided Deep Information Bottleneck (MSDIB), which extracts discriminative features and completes clustering through information compression and retention as well as a self-guided mechanism.
Multi-Attribute Multi-Grained Adaptation of Pre-Trained Language Models for Text Understanding from Bayesian Perspective
You Zhang (Yunnan University), Xuejie Zhang (Yunnan University)
Domain AdaptationKnowledge DistillationTransformerLarge Language ModelSupervised Fine-TuningText
🎯 What it does: A multi-attribute multi-granularity adaptation framework M2A is proposed, which integrates IID and non-IID features of pre-trained language models (PLM) from a Bayesian perspective, and achieves adaptive enhancement of text understanding tasks through efficient fine-tuning using parameters like LoRA.
Multi-axis Prompt and Multi-dimension Fusion Network for All-in-one Weather-degraded Image Restoration
Yuanbo Wen (Chang'an University), Ting Chen (Australian National University)
RestorationTransformerPrompt EngineeringImage
🎯 What it does: A multi-axis prompt and multi-dimensional fusion network, MPMF-Net, is proposed for the one-time removal of various weather noise in traffic images.
Multi-Branch Self-Drafting for LLM Inference Acceleration
Zipeng Gao (University of Science and Technology of China), Enhong Chen (University of Science and Technology of China)
OptimizationComputational EfficiencyTransformerLarge Language ModelTextBenchmark
🎯 What it does: Proposes the Self-Draft method, which utilizes the multi-branch reasoning of LLM itself to generate and validate drafts in parallel during a single forward pass, thereby accelerating large language model inference.
Multi-clue Consistency Learning to Bridge Gaps Between General and Oriented Object in Semi-supervised Detection
Chenxu Wang (Nanjing University of Science and Technology), Zhen Cui (Nanjing University of Science and Technology)
Object DetectionImage
🎯 What it does: This paper proposes a Multi-Channel Consistency Learning framework (MCL) aimed at addressing the issue of label assignment and confidence inconsistency in semi-supervised directed object detection.
Multi-concept Model Immunization through Differentiable Model Merging
Amber Yijia Zheng (Purdue University), Raymond A. Yeh (Purdue University)
GenerationData SynthesisOptimizationAdversarial AttackDiffusion modelImage
🎯 What it does: This study investigates an immune technology for multi-concept text-to-image models and proposes the MIMA algorithm, which enables the model to resist adaptive attacks from various harmful concepts before deployment.
Multi-Edge Reinforced Collaborative Data Acquisition for Continuous Video Analytics by Prioritizing Quality over Quantity
Lei Zhang (Nanjing University of Science and Technology), Huaizheng Zhang (Nanyang Technological University)
ClassificationObject DetectionFederated LearningTransformerReinforcement LearningVideo
🎯 What it does: This work proposes a multi-edge reinforcement collaborative video collection framework (AVA), which uses reinforcement learning to automatically select the most informative and non-redundant video frames from multiple edge cameras, significantly reducing the labeling and training costs of continuous learning.
Multi-fingered Hand Grasps with Visuo-Tactile Fusion via Multi-Agent Deep Reinforcement Learning
Peida Jia (Dalian University of Technology), Yi Sun (Dalian University of Technology)
Robotic IntelligenceReinforcement LearningMultimodality
🎯 What it does: The VT-MAGIC framework is proposed, which implements visual-tactile fusion control at the single finger level using multi-agent deep reinforcement learning, achieving stable grasping with a multi-fingered hand.
Multi-Focus Image Fusion via Explicit Defocus Blur Modelling
Yuhui Quan (South China University of Technology), Hui Ji (National University of Singapore)
Image TranslationRestorationConvolutional Neural NetworkRecurrent Neural NetworkImageBiomedical Data
🎯 What it does: This paper proposes a multi-focus image fusion network named DMANet, which generates pixel-level defocus descriptors and initial focused images through an explicit defocus blur model, and based on these results, generates soft decision maps and an uncertainty-aware fusion module, ultimately producing a panoramic focused image.
Multi-Frame Deformable Look-Up Table for Compressed Video Quality Enhancement
Gang He (Xidian University), Yunsong Li (Xidian University)
RestorationCompressionConvolutional Neural NetworkVideo
🎯 What it does: A multi-frame deformable lookup table (LUT) structure is proposed for video quality enhancement, which can significantly suppress the blurriness, ringing, and block effects caused by compression while maintaining real-time performance.
Multi-Grained Query-Guided Set Prediction Network for Grounded Multimodal Named Entity Recognition
Jielong Tang (Sun Yat-sen University), Jian Yin (Sun Yat-sen University)
RecognitionTransformerTextMultimodality
🎯 What it does: A unified framework called MQSPN based on multi-granularity query sets and set prediction is proposed to address the Grounded Multimodal Named Entity Recognition (GMNER) task based on image-text pairs.
Multi-Granular Multimodal Clue Fusion for Meme Understanding
Li Zheng (Wuhan University), Donghong Ji (Wuhan University)
ClassificationRecognitionConvolutional Neural NetworkContrastive LearningImageTextMultimodality
🎯 What it does: A multi-granularity multi-modal clue fusion model (MGMCF) is proposed to improve the multi-modal meme understanding (MMU) task.
Multi-Granularity Open Intent Classification via Adaptive Granular-Ball Decision Boundary
Yanhua Li (Southwestern University of Finance and Economics), Tianrui Li (Southwest Jiaotong University)
ClassificationRepresentation LearningTransformerSupervised Fine-TuningText
🎯 What it does: A multi-granularity open intent classification method MOGB is proposed, which combines adaptive granularity spherical clustering and the nearest sub-center classifier for hierarchical representation learning, and constructs multi-granularity decision boundaries to distinguish known intents from unknown intents.
Multi-Granularity Video Object Segmentation
Sangbeom Lim (Korea University), Seungryong Kim (KAIST)
Object TrackingSegmentationRecurrent Neural NetworkOptical FlowVideo
🎯 What it does: A multi-granularity video object segmentation dataset MUG-VOS is proposed, and a memory-based mask propagation model MMPM is trained and evaluated on it.
Multi-Instance Multi-Label Classification from Crowdsourced Labels
Ziquan Wang (Zhejiang University), Haobo Wang (ByteDance)
ClassificationText
🎯 What it does: This paper proposes a crowdsourced label learning framework (CMIML) for the multi-instance multi-label classification (MIML) task, achieving robust training under noisy labels by constructing a cross-label transition matrix and deriving an unbiased risk estimator.
Multi-Label Few-Shot Image Classification via Pairwise Feature Augmentation and Flexible Prompt Learning
Han Liu (Dalian University of Technology), Hong Yu (Dalian University of Technology)
ClassificationPrompt EngineeringImage
🎯 What it does: A multi-label few-shot image classification framework based on paired feature enhancement and flexible prompt learning is proposed, integrating prototypes generated from image region features with dynamically learned text prototypes.
Multi-Label Ranking Loss Minimization for Matrix Completion
Jiaxuan Li (Xi'an Jiaotong University), Jiayin Wang (Xi'an Jiaotong University)
Recommendation SystemDrug DiscoveryTabular
🎯 What it does: A matrix completion method based on multi-label ranking loss minimization (MLRM) is proposed, which uses the relative ranking among column vectors to constrain the completion results, thereby addressing the ambiguity caused by the low-rank assumption.
Multi-label Self Knowledge Distillation
Xucong Wang (University of Science and Technology of China), Yang Wang (Suzhou Institute for Advanced Research, University of Science and Technology of China)
Object DetectionKnowledge DistillationImage
🎯 What it does: The paper proposes a multi-label self-distillation method called MSKD, which implements self-teacher distillation using three spatial decoupling mechanisms.
Multi-Level Optimal Transport for Universal Cross-Tokenizer Knowledge Distillation on Language Models
Xiao Cui (University of Science and Technology of China), Houqiang Li (University of Science and Technology of China)
Knowledge DistillationTransformerSupervised Fine-TuningText
🎯 What it does: A MultiLevel Optimal Transport (MultiLevelOT) framework is proposed for cross-vocabulary knowledge distillation, aligning the logit distributions of the teacher and student at both the word level and sentence level using optimal transport (OT);
Multi-modal and Multi-scale Spatial Environment Understanding for Immersive Visual Text-to-Speech
Rui Liu (Inner Mongolia University), Haizhou Li (Chinese University of Hong Kong)
GenerationData SynthesisTransformerImageTextMultimodalityAudio
🎯 What it does: By combining RGB and Depth visual modalities and introducing a multi-scale local-global spatial understanding mechanism, the M2SE-VTTS model is proposed to achieve immersive visual text-to-speech (VTTS) synthesis based on environmental images.
Multi-modal Deepfake Detection via Multi-task Audio-Visual Prompt Learning
Hui Miao (Beihang University), Yunhong Wang (Beihang University)
Anomaly DetectionTransformerPrompt EngineeringVideoMultimodalityAudio
🎯 What it does: A multimodal deepfake video detection framework based on multi-task audio-video prompt learning is proposed, which can simultaneously utilize visual and audio information for detection.
Multi-Modal Grounded Planning and Efficient Replanning for Learning Embodied Agents with a Few Examples
Taewoong Kim (Seoul National University), Jonghyun Choi (Seoul National University)
OptimizationExplainability and InterpretabilityComputational EfficiencyRobotic IntelligenceTransformerLarge Language ModelTextMultimodalityBenchmarkRetrieval-Augmented Generation
🎯 What it does: This paper proposes FLARE, which combines a multimodal planner and environment-adaptive replanning to achieve interpretable task planning with a small amount of labeled data, enhancing the execution effectiveness of robots in the environment.
Multi-Modal Latent Variables for Cross-Individual Primary Visual Cortex Modeling and Analysis
Yu Zhu (Institute of Automation Chinese Academy of Sciences), Tiejun Huang (Beijing Academy of Artificial Intelligence)
Explainability and InterpretabilityRepresentation LearningAuto EncoderVideoMultimodality
🎯 What it does: Designed and implemented a multimodal identifiable variational autoencoder (miVAE) that maps V1 neural activity and visual stimuli to a unified latent variable space, and explains the source of latent variables through score attribution analysis.
Multi-Modal Recommendation Unlearning for Legal, Licensing, and Modality Constraints
Yash Sinha (National University of Singapore), Mohan Kankanhalli (National University of Singapore)
Recommendation SystemGraph Neural NetworkContrastive LearningMultimodality
🎯 What it does: This paper proposes MMRecUn, a no-learning (unlearning) framework for multimodal recommendation systems (MMRS) that allows for selective forgetting of user and item interactions.
Multi-Objective Evolution of Heuristic Using Large Language Model
Shunyu Yao (City University of Hong Kong), Qingfu Zhang (City University of Hong Kong)
OptimizationTransformerLarge Language ModelTabular
🎯 What it does: The MEoH framework is proposed, which combines large language models (LLM) with multi-objective evolutionary algorithms to automatically generate a set of non-dominated heuristic algorithms that satisfy multiple objectives (optimal gap and execution efficiency).
Multi-Objective Molecular Design Through Learning Latent Pareto Set
Yiping Liu (Hunan University), Hisao Ishibuchi (Southern University of Science and Technology)
OptimizationDrug DiscoveryAuto EncoderGraph
🎯 What it does: A multi-objective molecular design framework named MLPS is proposed, which utilizes an encoder-decoder to map discrete chemical space to continuous latent space, and combines local Bayesian optimization with a global Pareto set learning model to achieve high-quality molecular generation for multiple objectives (such as QED, SA, GSK3β, JNK3) as well as dual objectives of antifungal/toxicity.
Multi-Pair Temporal Sentence Grounding via Multi-Thread Knowledge Transfer Network
Xiang Fang (Huazhong University of Science and Technology), Beibei Li (Sichuan University)
RetrievalConvolutional Neural NetworkTransformerContrastive LearningVideoText
🎯 What it does: Designed and implemented a multi-pair TSG (multi-video-query joint training) framework MKTN, utilizing mechanisms such as cross-sentence semantic mining, cross-modal comparison, and prototype alignment to achieve efficient and accurate video-sentence alignment.
Multi-Perspective Consolidation Enhanced Cognitive Diagnosis via Conditional Diffusion Model
Guanhao Zhao (University of Science and Technology of China), Enhong Chen (University of Science and Technology of China)
TransformerDiffusion modelSequential
🎯 What it does: A multi-perspective integrated cognitive diagnosis framework (DMC-CDM) is proposed, which accurately captures cognitive states from a single perspective through a semantic extractor and a conditional diffusion model, and achieves information integration by maximizing mutual information across multiple perspectives.
Multi-Reference Preference Optimization for Large Language Models
Hung Le (Applied AI Institute), Svetha Venkatesh (Deakin University)
Recommendation SystemOptimizationTransformerLarge Language ModelReinforcement LearningText
🎯 What it does: A framework for Direct Preference Optimization using Multiple Reference Models (MRPO) is proposed to enhance the performance of large language models in various preference learning tasks.
Multi-Robot Task Allocation Using Global Games with Negative Feedback: The Colony Maintenance Problem
Logan E. Beaver (Old Dominion University)
OptimizationRobotic IntelligenceReinforcement LearningTabular
🎯 What it does: A global game mechanism based on negative feedback is proposed for the multi-robot collaborative maintenance of energy supply in community task allocation problems.
Multi-scale Activation, Refinement, and Aggregation: Exploring Diverse Cues for Fine-Grained Bird Recognition
Zhicheng Zhang (Nanjing University of Science and Technology), Jinhui Tang (Nanjing University of Science and Technology)
RecognitionTransformerContrastive LearningImage
🎯 What it does: A multi-scale multi-clue modeling (MDCM) framework is proposed, specifically targeting fine-grained bird recognition tasks.