AAAI 2025 Papers — Page 25
AAAI Conference on Artificial Intelligence · 3028 papers
S³-Mamba: Small-Size-Sensitive Mamba for Lesion Segmentation
Gui Wang (Shenzhen University), Linlin Shen (Shenzhen University)
SegmentationConvolutional Neural NetworkImageBiomedical DataMagnetic Resonance Imaging
🎯 What it does: A S³-Mamba model aimed at small lesion segmentation is proposed, specifically enhancing the recognition and segmentation capabilities of small lesions in medical images.
SADBA: Self-Adaptive Distributed Backdoor Attack Against Federated Learning
Jun Feng (Huazhong University of Science and Technology), Bocheng Ren (Hainan University)
Federated LearningAdversarial AttackImage
🎯 What it does: Proposes a Self-Adaptive Distributed Backdoor Attack (SADBA) that enhances the success rate of backdoor attacks in a federated learning environment while significantly reducing the proportion of malicious clients.
Safe Online Convex Optimization with Heavy-Tailed Observation Noises
Yunhao Yang (Zhejiang University), Yuanyu Wan (Zhejiang University)
Optimization
🎯 What it does: A safe online convex optimization algorithm is proposed under unknown linear safety constraints, where the observation noise follows a heavy-tailed distribution. This algorithm ensures that all decisions meet safety constraints and achieves optimal regret with high probability.
Safe Planner: Empowering Safety Awareness in Large Pre-Trained Models for Robot Task Planning
Siyuan Li (Harbin Institute of Technology), Xun Wang (Intelligent Science and Technology Academy Limited of CASIC)
Safty and PrivacyRobotic IntelligenceTransformerLarge Language ModelVision Language ModelMultimodalitySequential
🎯 What it does: The paper proposes a safety planning framework called Safe Planner, which utilizes large pre-trained models (LLM/VLM) in conjunction with a learnable safety prediction module to generate executable and safe robotic task planning.
SAFIRE: Segment Any Forged Image Region
Myung-Joon Kwon (Korea Advanced Institute of Science and Technology), Changick Kim (NAVER WEBTOON AI)
SegmentationContrastive LearningImage
🎯 What it does: We propose SAFIRE, a point-guided multi-source image segmentation method designed to achieve both counterfeit localization and multi-source differentiation simultaneously.
SAIL: Sample-Centric In-Context Learning for Document Information Extraction
Jinyu Zhang (Shanghai Jiao Tong University), Xinyi Le (Shanghai Jiao Tong University)
Data-Centric LearningTransformerLarge Language ModelPrompt EngineeringText
🎯 What it does: A sample-centered context learning framework (SAIL) is proposed for zero-shot document information extraction.
Salient Frequency-aware Exemplar Compression for Resource-constrained Online Continual Learning
Junsu Kim (Korea University), Suhyun Kim (Kyung Hee University)
ClassificationCompressionImage
🎯 What it does: This paper proposes a frequency saliency-aware example compression framework (SFEC) based on JPEG, which stores more high-quality examples with extremely low computational overhead in online incremental learning and further alleviates compression distortion through compressed sensing buffer management.
SalM²: An Extremely Lightweight Saliency Mamba Model for Real-Time Cognitive Awareness of Driver Attention
Chunyu Zhao (Southwest Jiaotong University), Tao Deng (Southwest Jiaotong University)
Autonomous DrivingContrastive LearningImage
🎯 What it does: A lightweight SalM2 driver attention prediction model has been developed, capable of capturing attention distribution in driving scenes in real-time.
SAM-Aware Graph Prompt Reasoning Network for Cross-Domain Few-Shot Segmentation
Shi-Feng Peng (Nanjing University of Science and Technology), Guo-Sen Xie (Southeast University)
SegmentationGraph Neural NetworkLarge Language ModelImageBenchmark
🎯 What it does: Proposes the SAM-aware Graph Prompt Reasoning Network (GPRN), which generates visual prompts through SAM, utilizes a graph attention network to achieve global semantic consistency among prompts, and employs adaptive point selection for refined segmentation during testing.
Sample Complexity of Linear Regression Models for Opinion Formation in Networks
Haolin Liu (University of Virginia), Haifeng Xu (University of Chicago)
Graph
🎯 What it does: This paper studies the minimum total sample size required for a linear regression model to achieve the desired error through opinion formation games in social networks, and analyzes the impact of different network structures on sample complexity.
Sample-aware Adaptive Structured Pruning for Large Language Models
Jun Kong (Yunnan University), Xuejie Zhang (Yunnan University)
TransformerLarge Language ModelSupervised Fine-TuningText
🎯 What it does: This paper proposes a sample-aware adaptive structural pruning framework for large language models called AdaPruner, which achieves more efficient structural pruning by adaptively searching for the best calibration data and importance estimation metrics.
SAP: Corrective Machine Unlearning with Scaled Activation Projection for Label Noise Robustness
Sangamesh Kodge (Purdue University), Kaushik Roy (Purdue University)
ClassificationData-Centric LearningConvolutional Neural NetworkTransformerImage
🎯 What it does: This paper proposes a single-step correction machine learning algorithm based on Singular Value Decomposition (SVD) called SAP, which uses a small number of low cross-entropy loss samples to estimate reliable samples and projects the model weights into a reliable activation space, thereby eliminating the impact of label noise.
SatCLIP: Global, General-Purpose Location Embeddings with Satellite Imagery
Konstantin Klemmer (Microsoft Research), Marc Rußwurm (Wageningen University and Research)
Representation LearningTransformerContrastive LearningImage
🎯 What it does: This paper proposes SatCLIP, a globally universal location encoder trained through contrastive learning by aligning satellite images with latitude and longitude coordinates, and fine-tuned for various geographic prediction tasks.
SAUGE: Taming SAM for Uncertainty-Aligned Multi-Granularity Edge Detection
Xing Liufu (Sun Yat-sen University), Jian-Fang Hu (Sun Yat-sen University)
Object DetectionSegmentationConvolutional Neural NetworkSupervised Fine-TuningImage
🎯 What it does: We propose SAUGE, an edge detection framework that utilizes intermediate features from the Segment Anything Model (SAM) and generates multi-granularity edge maps through a lightweight side transfer network, allowing for flexible control over edge thickness as needed.
Scaffold-BPE: Enhancing Byte Pair Encoding for Large Language Models with Simple and Effective Scaffold Token Removal
Haoran Lian (Beihang University), Guiguang Ding (Tsinghua University)
TransformerLarge Language ModelText
🎯 What it does: This paper proposes Scaffold-BPE, an improved method for dynamically removing low-frequency scaffold words during the BPE training and encoding process.
Scalable Acceleration for Classification-Based Derivative-Free Optimization
Tianyi Han (Beijing Supreium Technology), Yuan Jin (Beijing Supreium Technology)
OptimizationHyperparameter SearchTabular
🎯 What it does: This paper studies a classification-based gradient-free optimization algorithm framework and proposes a new RACE-CARS algorithm, which accelerates the search through random coordinate shrinking and region reduction.
Scalable Decentralized Algorithms for Online Personalized Mean Estimation
Franco Galante (Politecnico di Torino), Emilio Leonardi (Politecnico di Torino)
Federated LearningComputational EfficiencyImage
🎯 What it does: Two scalable decentralized online mean estimation algorithms (B‑ColME and C‑ColME) are proposed, which estimate the mean with communication limited to a finite number of neighbors under multi-dimensional distributions.
Scalable Federated One-Step Multi-View Clustering with Tensorized Regularization
Wei Feng (Xi'an Jiaotong University), Zheng Yan (Xidian University)
Federated LearningSafty and PrivacyMultimodality
🎯 What it does: A federated multi-view clustering framework SFOMVC-TR based on anchor graphs and tensor regularization is proposed, achieving one-stage clustering while ensuring privacy.
Scalable Knowledge Refactoring Using Constrained Optimisation
Minghao Liu (University of Oxford), Andrew Cropper (University of Oxford)
CompressionOptimizationTabularAlzheimer's Disease
🎯 What it does: This paper proposes a knowledge reconstruction method based on constraint optimization, aimed at compressing the size of logic programs.
Scalable One-Pass Incomplete Multi-View Clustering by Aligning Anchors
Yalan Qin (Shanghai University), Xinpeng Zhang (Shanghai University)
Computational EfficiencyImage
🎯 What it does: SOME-AS is proposed, a scalable one-time incomplete multi-view clustering method achieved through anchor point alignment and subspace learning.
Scalable Quantum-Inspired Optimization Through Dynamic Qubit Compression
Co Tran (University of Texas), Thang N Dinh (Virginia Commonwealth University)
OptimizationGraph Neural NetworkGraph
🎯 What it does: A dynamic compression framework called GRANITE based on graph neural networks is proposed to compress large-scale Ising models for adaptation to quantum annealers with a limited number of qubits.
Scalable Surrogate Verification of Image-Based Neural Network Control Systems Using Composition and Unrolling
Feiyang Cai (Clemson University), Stanley Bak (Stony Brook University)
OptimizationConvolutional Neural NetworkTransformerGenerative Adversarial NetworkImage
🎯 What it does: This paper proposes a method for proxy verification of perception systems using Conditional Generative Adversarial Networks (cGAN), significantly reducing the over-approximation of reachable sets through a combination of system dynamics and networks with multi-step unfolding, thereby enhancing the feasibility of safety verification for image-driven neural network control systems.
Scalable Trajectory-User Linking with Dual-Stream Representation Networks
Hao Zhang (Ocean University of China), Yanwei Yu (Ocean University of China)
Recurrent Neural NetworkContrastive LearningTime SeriesSequential
🎯 What it does: This paper proposes ScaleTUL, a scalable trajectory-user linking framework that utilizes a dual-stream encoder and spatiotemporal enhancement to address the large-scale trajectory matching problem.
ScaleMatch: Multi-scale Consistency Enhancement for Semi-supervised Semantic Segmentation
Liang Lv (Wuhan University), Lefei Zhang (Wuhan University)
SegmentationConvolutional Neural NetworkSupervised Fine-TuningImage
🎯 What it does: The ScaleMatch framework is proposed to enhance the scale invariance and pseudo-label quality of semi-supervised semantic segmentation models under limited labeled data through Cross-Scale Interaction Fusion (CIF) and Intra-Scale Consistency Learning (ISVC, FSVC).
ScaleOT: Privacy-utility-scalable Offsite-tuning with Dynamic LayerReplace and Selective Rank Compression
Kai Yao (Ant Group), Jianke Zhu (Zhejiang University)
OptimizationSafty and PrivacyTransformerLarge Language ModelReinforcement LearningText
🎯 What it does: This paper proposes ScaleOT, an offline tuning framework that combines Dynamic Layer Replace and Selective Rank Compression, capable of generating low-compression simulators of different scales to achieve a balance between model privacy and fine-tuning performance.
Scaling Combinatorial Optimization Neural Improvement Heuristics with Online Search and Adaptation
Federico Julian Camerota Verdù (University of Trieste), Luca Bortolussi (University of Trieste)
OptimizationTransformerReinforcement LearningTabular
🎯 What it does: An improved heuristic based on Limited Rollback Beam Search (LRBS) is proposed to enhance the reasoning performance of deep reinforcement learning models in combinatorial optimization.
Scaling Diffusion Mamba with Bidirectional SSMs for Efficient 3D Shape Generation
Shentong Mo (Carnegie Mellon University)
GenerationData SynthesisDiffusion modelPoint CloudMesh
🎯 What it does: A Diffusion Mamba (DiM-3D) architecture based on the Mamba state space model is proposed for high-resolution 3D shape (voxel/point cloud) generation.
SCALM: Detecting Bad Practices in Smart Contracts Through LLMs
Zongwei Li (Hainan University), Xin Wang (Hainan University)
TransformerLarge Language ModelPrompt EngineeringTextRetrieval-Augmented Generation
🎯 What it does: Proposed and implemented the SCALM framework, which utilizes large language models to detect bad practices in smart contracts.
SCANS: Mitigating the Exaggerated Safety for LLMs via Safety-Conscious Activation Steering
Zouying Cao (Shanghai Jiao Tong University), Hai Zhao (Shanghai Jiao Tong University)
Safty and PrivacyLarge Language ModelPrompt EngineeringText
🎯 What it does: A security awareness control method based on activation layer driving (SCANS) is proposed, which mitigates the misrejection of legitimate queries by extracting rejection behavior vectors and performing activation shifts in security-critical layers.
SCCS: Deep Neural Spectral Clustering for Self-Supervised Subcellular Structure Segmentation
Jimao Jiang (Peking University), Yuru Pei (Peking University)
SegmentationGraph Neural NetworkContrastive LearningImageBiomedical Data
🎯 What it does: A lightweight graph neural network framework called SCCS based on self-supervised deep spectral clustering is proposed for end-to-end cellular substructure segmentation on superpixel graphs.
Scenario-Based Robust Optimization of Tree Structures
Spyros Angelopoulos (Sorbonne University), Georgii Melidi
OptimizationTabular
🎯 What it does: The study constructs robust binary search trees and Huffman trees under multi-scenario frequency distributions, and proposes performance metrics such as competitive ratio and regret.
Scene Graph-Grounded Image Generation
Fuyun Wang (Nanjing University of Science and Technology), Zhen Cui (Nanjing University of Science and Technology)
GenerationData SynthesisGraph Neural NetworkDiffusion modelAuto EncoderContrastive LearningImage
🎯 What it does: A scene graph-based image generation framework SGG-IG is designed and implemented, directly aligning scene graphs with generated images without an intermediate layout, integrating modules such as mask self-supervision, spatial contrastive loss, and image-scene alignment.
SceneX: Procedural Controllable Large-Scale Scene Generation
Mengqi Zhou (University of Chinese Academy of Sciences), Zhaoxiang Zhang (University of Chinese Academy of Sciences)
GenerationTransformerLarge Language ModelText
🎯 What it does: Developed the SceneX framework, achieving automated large-scale scene generation and editing based on LLM in Blender.
ScholarGEC: Enhancing Controllability of Large Language Model for Chinese Academic Grammatical Error Correction
Zixiao Kong (University of Science and Technology of China), Yu Su (Hefei Normal University)
TransformerLarge Language ModelSupervised Fine-TuningPrompt EngineeringText
🎯 What it does: This study proposes the ScholarGEC framework, which enhances the controllability and accuracy of LLM in grammatical error correction of Chinese academic papers by utilizing knowledge prefixes, a detect-correct structure, and self-reflection techniques.
SCKD: Semi-Supervised Cross-Modality Knowledge Distillation for 4D Radar Object Detection
Ruoyu Xu (Zhejiang University), Eryun Liu (Zhejiang University)
Object DetectionKnowledge DistillationMultimodalityPoint Cloud
🎯 What it does: This paper proposes a 4D radar 3D object detection method based on semi-supervised cross-modal knowledge distillation, called SCKD.
SCOPE: Sign Language Contextual Processing with Embedding from LLMs
Yuqi Liu (ShanghaiTech University), Lan Xu (ShanghaiTech University)
RecognitionImage TranslationTransformerLarge Language ModelSupervised Fine-TuningVideoText
🎯 What it does: This paper proposes a context-aware sign language recognition and translation framework called SCOPE based on LLM embeddings, and contributes the first large-scale Chinese sign language dialogue dataset.
ScoreNet: Consistency-driven Framework with Multi-side Information Fusion for Session-based Recommendation
Piao Tong (University of Electronic Science and Technology of China), Tian Lan (Byte Dance)
Recommendation SystemTransformerSequential
🎯 What it does: Proposes the ScoreNet framework, which combines the MPRE and PECS modules for multi-side information fusion, improving interest modeling and score consistency in session-based recommendations.
SCott: Accelerating Diffusion Models with Stochastic Consistency Distillation
Hongjian Liu (OPPO AI Center), Zheng-Jun Zha (University of Science and Technology of China)
GenerationData SynthesisDiffusion modelGenerative Adversarial NetworkImageStochastic Differential Equation
🎯 What it does: By combining a stochastic differential equation (SDE) solver with consistency distillation, the SCott model is proposed to accelerate the sampling of text-to-image diffusion models, generating high-quality images in just 2 to 4 steps.
ScreenMark: Watermarking Arbitrary Visual Content on Screen
Xiujian Liang (Fudan University), Zhenxing Qian (Fudan University)
GenerationSafty and PrivacyDiffusion modelImageMultimodality
🎯 What it does: This paper proposes ScreenMark, a three-stage deep learning watermarking scheme for arbitrary visual screen content, achieving real-time protection of screen content.
SdalsNet: Self-Distilled Attention Localization and Shift Network for Unsupervised Camouflaged Object Detection
Peiyao Shou (Hangzhou Dianzi University), Chenggang Yan (Hangzhou Dianzi University)
Object DetectionSegmentationKnowledge DistillationTransformerImage
🎯 What it does: This paper proposes a fully unsupervised camouflage object detection method called SdalsNet, which utilizes self-distillation attention localization and transfer to learn target features and directly obtain segmentation results from the attention map.
Search Strategy Generation for Branch and Bound Using Genetic Programming
Gwen Maudet (University of Luxembourg), Grégoire Danoy (University of Luxembourg)
Optimization
🎯 What it does: This paper proposes a search strategy generation method based on genetic programming, GP2S, for the automated generation of Branch-and-Bound (BB) search strategies.
Searching for and Avoiding Hidden Sets Using Queries with Local Feedback
Tomasz Jurdzinski (University of Wroclaw), Dariusz R. Kowalski (Augusta University)
🎯 What it does: This study proposes an enhanced group testing method that aims to discover elements of hidden sets by introducing a local feedback framework and avoiding queries from interference sets.
SEAS: Self-Evolving Adversarial Safety Optimization for Large Language Models
Muxi Diao (Beijing University of Posts and Telecommunications), Weiran Xu (Meituan)
OptimizationSafty and PrivacyAdversarial AttackTransformerLarge Language ModelSupervised Fine-TuningReinforcement LearningText
🎯 What it does: An adaptive red team framework SEAS is proposed, which generates diverse adversarial prompts through bidirectional iteration between the red team model and the target model, enhancing the security of the target model.
SECodec: Structural Entropy-based Compressive Speech Representation Codec for Speech Language Models
Linqin Wang (Kunming University of Science and Technology), Ling Dong (Kunming University of Science and Technology)
GenerationCompressionTransformerGenerative Adversarial NetworkAudio
🎯 What it does: This paper proposes a speech discretization encoder SECodec based on Structural Entropy (SE) and a speech language model SESLM driven by Structural Entropy, which can automatically determine the codebook size and reduce Euclidean distance distortion through a 2D structural entropy quantization method, thereby enhancing speech reconstruction and zero-shot TTS performance.
Security Attacks on LLM-based Code Completion Tools
Wen Cheng (Nanjing University), Wei Wang (Nanjing University)
Safty and PrivacyAdversarial AttackAI Code AssistantTransformerLarge Language ModelPrompt EngineeringText
🎯 What it does: This paper studies the security vulnerabilities of large language model-driven code completion tools (LCCT) and proposes and validates two types of attacks against LCCT: jailbreak attacks based on context information aggregation and hierarchical code exploitation, as well as code-based training data extraction attacks.
See Through Their Minds: Learning Transferable Brain Decoding Models from Cross-Subject fMRI
Yulong Liu (Xi'an Jiaotong University), Nanning Zheng (Xi'an Jiaotong University)
ClassificationRecognitionGenerationRetrievalDiffusion modelContrastive LearningImageTextMagnetic Resonance Imaging
🎯 What it does: This study investigates cross-subject fMRI brain decoding models, using shallow adapters to map fMRI data from different subjects into a unified representational space, and then performing high-level semantic recognition and low-level pixel-level reconstruction through a shared deep decoding module.
SeeDiff: Off-the-Shelf Seeded Mask Generation from Diffusion Models
Joon Hyun Park (Hanyang University), Sungyong Baik (Hanyang University)
SegmentationData SynthesisDiffusion modelImage
🎯 What it does: This paper proposes a training-free and prompt-free offline framework called SeeDiff, which utilizes the attention maps from Stable Diffusion to generate high-quality pixel-level segmentation masks and can directly synthesize semantic segmentation datasets.
Seeing Beyond Noise: Joint Graph Structure Evaluation and Denoising for Multimodal Recommendation
Yuxin Qi (Shanghai Jiao Tong University), Jianhua Li (Shanghai Jiao Tong University)
Recommendation SystemGraph Neural NetworkContrastive LearningMultimodalityGraph
🎯 What it does: A multimodal recommendation framework called EVEN is proposed, which simultaneously evaluates and denoises semantic priors and user interaction noise.
Seeing Your Speech Style: A Novel Zero-Shot Identity-Disentanglement Face-based Voice Conversion
Yan Rong (Hong Kong University of Science and Technology), Li Liu (Hong Kong University of Science and Technology)
GenerationData SynthesisTransformerContrastive LearningTextAudio
🎯 What it does: What was done: Proposed the ID-FaceVC framework, achieving zero-shot voice conversion, supporting audio or text input, and allowing control over emotion and speech rate.
SeFAR: Semi-supervised Fine-grained Action Recognition with Temporal Perturbation and Learning Stabilization
Yongle Huang (Northwestern Polytechnical University), Dian Shao (Northwestern Polytechnical University)
RecognitionTransformerLarge Language ModelVideoMultimodality
🎯 What it does: The SeFAR framework is proposed, which enhances performance in semi-supervised fine-grained action recognition through dual-layer temporal element modeling, moderate temporal perturbation enhancement, and adaptive regulation.
Seg2Box: 3D Object Detection by Point-Wise Semantics Supervision
Maoji Zheng (Xiamen University), Cheng Wang (Xiamen University)
Object DetectionSegmentationAutonomous DrivingPoint Cloud
🎯 What it does: This paper proposes a framework called Seg2Box for 3D object detection that solely utilizes semantic labels. It first generates high-quality pseudo boxes through multi-frame multi-scale clustering, and then continuously mines unlabeled instances and refines the pseudo boxes using self-training loops.
SegFace: Face Segmentation of Long-Tail Classes
Kartik Narayan (Johns Hopkins University), Vishal M. Patel (Johns Hopkins University)
SegmentationTransformerImage
🎯 What it does: This paper proposes SegFace, a face segmentation method that utilizes a lightweight transformer decoder and learnable class-specific tokens, specifically aimed at improving long-tail categories.
Segment Any 3D Gaussians
Jiazhong Cen (Shanghai Jiao Tong University), Qi Tian (Huawei Technologies Co Ltd)
SegmentationKnowledge DistillationContrastive LearningGaussian SplattingPoint Cloud
🎯 What it does: Developed the SAGA method to achieve promptable multi-scale 3D segmentation within a 3D Gaussian expansion framework.
Selective Forgetting: Advancing Machine Unlearning Techniques and Evaluation in Language Models
Lingzhi Wang (Harbin Institute of Technology), Georg Gottlob (University of Calabria)
Data-Centric LearningTransformerLarge Language ModelText
🎯 What it does: The SEUL method is proposed to achieve selective and fine-grained forgetting in language models, and sensitive information forgetting evaluation metrics S-EL, S-MA, as well as an online/offline sensitive span automatic labeling process are designed.
Selective Uncertainty Propagation in Offline RL
Sanath Kumar Krishnamurthy (Meta), Anshuka Rangi (Indian Institute of Technology Bombay)
Reinforcement Learning
🎯 What it does: This paper proposes a method for selective uncertainty propagation in offline reinforcement learning, constructing confidence intervals for estimating the treatment effect of each decision step, and based on this, designs an improved SPVI (Selective Pessimistic Value Iteration) algorithm.
Selective Visual Prompting in Vision Mamba
Yifeng Yao (Peking University), Jiahuan Zhou (Peking University)
ClassificationRecognitionTransformerPrompt EngineeringImage
🎯 What it does: A Selective Visual Prompting (SVP) method specifically designed for Vision Mamba is proposed and implemented for efficient fine-tuning.
SELF-[IN]CORRECT: LLMs Struggle with Discriminating Self-Generated Responses
Dongwei Jiang (Johns Hopkins University), Daniel Khashabi (Johns Hopkins University)
GenerationTransformerLarge Language ModelPrompt EngineeringTextChain-of-Thought
🎯 What it does: This paper proposes a unified framework that can compare and evaluate the generative and discriminative (self-correction) capabilities of LLMs on the same tasks, and conducts experimental validation on the 'SELF‑[IN]CORRECT' hypothesis.
Self-attention-based Diffusion Model for Time-series Imputation in Partial Blackout Scenarios
Mohammad Rafid Ul Islam (Oregon State University), Alan Fern (Oregon State University)
Diffusion modelScore-based ModelTime Series
🎯 What it does: This paper proposes a self-attention-based diffusion model SADI for high-quality imputation of missing values in multivariate time series under the 'partial blackout' missing pattern.
Self-Attentive Spatio-Temporal Calibration for Precise Intermediate Layer Matching in ANN-to-SNN Distillation
Di Hong (Zhejiang University), Yueming Wang (Zhejiang University)
Knowledge DistillationSpiking Neural NetworkImage
🎯 What it does: This paper studies a Self-Attention Spatio-Temporal Calibration (SASTC) method to improve ANN-to-SNN knowledge distillation.
Self-Corrected Flow Distillation for Consistent One-Step and Few-Step Image Generation
Quan Dao (Rutgers University), Anh Tran (VinAI Research)
GenerationData SynthesisKnowledge DistillationFlow-based ModelRectified FlowGenerative Adversarial NetworkImage
🎯 What it does: Distill the pre-trained flow matching teacher model into a model capable of achieving single-step and few-step consistent image generation.
Self-Correcting Robot Manipulation via Gaussian-Splatted Foresight
Shaohui Pan (South China University of Technology), Zhuliang Yu (South China University of Technology)
Robotic IntelligenceTransformerGaussian SplattingPoint Cloud
🎯 What it does: This paper proposes a robot operation self-correction method based on 3D Gaussian distribution foresight, utilizing Gaussian Splatting to model the current scene and predicting the future scene after executing actions through a prediction network; when the prediction significantly deviates from the actual observation, the robot rolls back to the previous action and makes small perturbations to achieve failure detection and recovery.
Self-Evolutionary Large Language Models Through Uncertainty-Enhanced Preference Optimization
Jianing Wang (Meituan), Peng Yan (Meituan)
OptimizationReinforcement Learning from Human FeedbackTransformerLarge Language ModelSupervised Fine-TuningReinforcement LearningText
🎯 What it does: A self-evolutionary preference optimization (UPO) framework based on uncertainty enhancement is proposed, which filters reliable preference data by estimating the uncertainty of generated response pairs and updates the LLM with this data in each iteration to improve the model's alignment with human preferences.
Self-Explainable Graph Transformer for Link Sign Prediction
Lu Li (Huazhong Agricultural University), Zeyu Zhang (Huazhong Agricultural University)
Recommendation SystemExplainability and InterpretabilityGraph Neural NetworkTransformerGraph
🎯 What it does: Developed a self-explanatory signature graph Transformer (SE-SGformer) that can predict link symbols and provide interpretable positive/negative neighbor information.
Self-Prompting Analogical Reasoning for UAV Object Detection
Nianxin Li (University of Electronic Science and Technology of China), Xiatian Zhu (University of Surrey)
Object DetectionVision Language ModelImage
🎯 What it does: The SPAR method is proposed, which combines self-prompting and analogical reasoning to achieve precise detection of small targets in drone images within the YOLOv8 framework.
Self-Supervised Collaborative Information Bottleneck for Text Readability Assessment
Jinshan Zeng (Jiangxi Normal University), Wenyan Xiao (Jiangxi University of Science and Technology)
ClassificationTransformerLarge Language ModelText
🎯 What it does: A hybrid readability assessment model SCIB-ARA based on self-supervised collaborative information bottleneck is proposed, which evaluates text readability by integrating multi-layer deep representations (word, sentence, document) with linguistic features.
Self-supervised Trusted Contrastive Multi-view Clustering with Uncertainty Refined
Shizhe Hu (Zhengzhou University), Yangdong Ye (Zhengzhou University)
Representation LearningContrastive LearningImage
🎯 What it does: A self-supervised trustworthy contrastive learning method is proposed for unsupervised multi-view clustering, improving feature representation and clustering results through evidence fusion and uncertainty learning.
Semantic Ambiguity Modeling and Propagation for Fine-Grained Visual Cross View Geo-Localization
Mingtao Feng (Xidian University), Ajmal Saeed Mian
RecognitionRetrievalConvolutional Neural NetworkContrastive LearningImage
🎯 What it does: This paper studies the semantic ambiguity of semi-positive sample generation in visual cross-view geographic localization, proposing an automatic balance between retrieval and offset regression joint training by modeling the uncertainty of offset regression and propagating it to the retrieval task.
Semantic Convergence: Harmonizing Recommender Systems via Two-Stage Alignment and Behavioral Semantic Tokenization
Guanghan Li (Meituan), Wei Lin (Meituan)
Recommendation SystemTransformerLarge Language ModelSupervised Fine-TuningTextSequential
🎯 What it does: This paper proposes a two-stage alignment framework that integrates large language models (LLM) with traditional recommendation systems, achieving semantic mapping and alignment of item IDs.
Semantic Enhanced Heterogeneous Hypergraph Network for Collaborative Filtering
Mingtao Xu (Huazhong University of Science and Technology), Hulong Wu (Shenzhen Yishi Huolala Technology Limited)
Recommendation SystemGraph Neural NetworkLarge Language ModelAuto EncoderTextGraph
🎯 What it does: The SEHHN model is proposed, which constructs a heterogeneous hypergraph using the semantic information of comments and item category information extracted by LLM, thereby enhancing the collaborative filtering recommendation effect.
Semantic Segmentation on Raindrop Degraded Images Using Two-Stage Dual Teacher-Student Learning
Xin Yang (National University of Singapore), Robby T. Tan (National University of Singapore)
SegmentationDomain AdaptationKnowledge DistillationConvolutional Neural NetworkImage
🎯 What it does: A dual teacher-student two-stage framework is proposed to separately address visual degradation and occlusion caused by raindrops, enhancing semantic segmentation performance.
Semantic-guided Masked Mutual Learning for Multi-modal Brain Tumor Segmentation with Arbitrary Missing Modalities
Guoyan Liang (Zhejiang University), Kai Chen (Gongli Hospital of Shanghai Pudong New Area)
SegmentationKnowledge DistillationConvolutional Neural NetworkTransformerMultimodalityBiomedical DataMagnetic Resonance Imaging
🎯 What it does: A semantic-guided mask mutual learning (SMML) framework is proposed to achieve brain tumor segmentation under the condition of any missing modality using a dual-branch mask mechanism.
Semi-IIN: Semi-Supervised Intra-Inter Modal Interaction Learning Network for Multimodal Sentiment Analysis
Jinhao Lin (South China University of Technology), Qi Liu (South China University of Technology)
ClassificationConvolutional Neural NetworkMultimodality
🎯 What it does: Proposes the Semi-IIN network, which combines self-training with unsupervised data to dynamically separate and select multimodal interaction information for sentiment analysis.
Semi-Implicit Neural Ordinary Differential Equations
Hong Zhang (Argonne National Laboratory), Romit Maulik (Pennsylvania State University)
GraphTime SeriesOrdinary Differential Equation
🎯 What it does: A semi-implicit neural ODE (SINODE) framework is proposed, utilizing linear-nonlinear partitioning and IMEX Runge-Kutta integration, and achieving reversible exact gradients through discrete adjoint differentiation.
Semi-supervised 3D Semantic Scene Completion with 2D Vision Foundation Model Guidance
Duc-Hai Pham (VinAI Research), Rang Nguyen (VinAI Research)
SegmentationDepth EstimationAutonomous DrivingTransformerPoint Cloud
🎯 What it does: A semi-supervised 3D semantic scene completion framework VFG-SSC is proposed, which utilizes 2D vision-based models to generate 3D priors, significantly reducing the reliance on large-scale 3D annotations.
Semi-Supervised Clustering Framework for Fine-grained Scene Graph Generation
Jiarui Yang (Fudan University), Liang Yang
Object DetectionGenerationTransformerContrastive LearningImage
🎯 What it does: A semi-supervised clustering framework SSC-SGG is proposed, which utilizes limited labeled data to mine low-frequency interaction relationships among unlabeled object pairs and generate pseudo-labels, enriching the training sample space.
Semi-supervised Infrared Small Target Detection with Thermodynamic-Inspired Uneven Perturbation and Confidence Adaptation
Mingjin Zhang (Xidian University), Jing Zhang (Wuhan University)
Object DetectionConvolutional Neural NetworkSupervised Fine-TuningImage
🎯 What it does: A semi-supervised learning-based infrared small target detection framework (S3D) is proposed, which can significantly improve detection performance in cases of scarce samples.
Semi-Supervised Multi-View Multi-Label Learning with View-Specific Transformer and Enhanced Pseudo-Label
Quanjiang Li (National University of Defense Technology), Feijiang Li (Shanxi University)
ClassificationTransformerGenerative Adversarial NetworkContrastive LearningMultimodality
🎯 What it does: A semi-supervised multi-view multi-label learning framework SMVTEP is designed, which integrates view-specific Transformers and enhanced pseudo-labels to improve classification performance under limited annotations.
Semi-Supervised Multimodal Classification Through Learning from Modal and Strategic Complementarities
Junchi Chen (Beihang University), Junfan Chen (Beihang University)
ClassificationTransformerContrastive LearningImageTextMultimodality
🎯 What it does: This paper studies image-text semi-supervised multimodal classification in scenarios with scarce labels and proposes the MSC framework.
Semi-Supervised Online Cross-Modal Hashing
Xiao Kang (Shandong University), Yilong Yin (Shandong University)
RetrievalOptimizationMultimodality
🎯 What it does: A semi-supervised online cross-modal hashing framework SSOCH is proposed, which can simultaneously handle labeled and unlabeled, paired and unpaired multimodal samples in data streams, and update the hashing function in real-time.
SemiDFL: A Semi-Supervised Paradigm for Decentralized Federated Learning
Xinyang Liu (Shenzhen Institute of Artificial Intelligence and Robotics for Society), Bo Liu
Federated LearningDiffusion modelImage
🎯 What it does: This paper proposes SemiDFL, a new framework for semi-supervised federated learning that operates without a central server, enabling collaborative training among clients with label scarcity and highly non-homogeneous distributions through model and data space consensus.
SemStereo: Semantic-Constrained Stereo Matching Network for Remote Sensing
Chen Chen (Aerospace Information Research Institute, Chinese Academy of Sciences), Xian Sun (Aerospace Information Research Institute, Chinese Academy of Sciences)
SegmentationDepth EstimationConvolutional Neural NetworkImage
🎯 What it does: A multi-task network SemStereo is designed to perform semantic segmentation and stereo matching simultaneously, enhancing the accuracy of both tasks through semantic constraints.
Sensing Surface Patches in Volume Rendering for Inferring Signed Distance Functions
Sijia Jiang (Wayne State University), Zhizhong Han (Wayne State University)
Depth EstimationOptimizationNeural Radiance FieldImage
🎯 What it does: Inferring the signed distance function (SDF) from multi-view RGB images through volumetric rendering, and explicitly applying surface constraints by stretching points on the zero level set to form surface patches, thereby improving geometric detail recovery.
Sentence-level Aggregation of Lexical Metrics Correlates Stronger with Human Judgements than Corpus-level Aggregation
Paulo Cavalin (IBM Research), Claudio Pinhanez (IBM Research)
Text
🎯 What it does: This paper compares the effects of Corpus-Level Aggregation (CLA) and Sentence-Level Aggregation (SLA) on lexical evaluation metrics (such as BLEU and chrF), finding that the latter has a higher correlation with human evaluations and neural metrics (COMET, BLEURT, BERTScore).
Separating the Wheat from the Chaff: Spatio-Temporal Transformer with View-interweaved Attention for Photon-Efficient Depth Sensing
Letian Yu (Dalian University of Technology), Xin Yang (Dalian University of Technology)
Depth EstimationTransformerImage
🎯 What it does: A 3D space-time Transformer (SPDS-STFormer) is proposed for single-photon depth perception, which reconstructs high-quality depth maps under low photon counts and low signal-to-noise ratio conditions through View Interleaved Attention (VIAM) and an adaptive weighting mechanism.
Sequence Accumulation and Beyond: Infinite Context Length on Single GPU and Large Clusters
Weigao Sun (Shanghai AI Laboratory), Xiaoyu Mo (Nanyang Technological University)
Recurrent Neural NetworkTransformerLarge Language ModelTextSequential
🎯 What it does: This paper proposes Sequence Accumulation (SA) and Sequence Accumulation with Pipeline Parallelism (SAPP), enabling linear sequence models to be trained with infinite context lengths on a single GPU or even large-scale clusters while maintaining constant memory.
Sequence Complementor: Complementing Transformers for Time Series Forecasting with Learnable Sequences
Xiwen Chen (Clemson University), Abolfazl Razi (Arizona State University)
TransformerTime SeriesSequential
🎯 What it does: A learnable sequence complementor is added to the Transformer model to enhance the representational richness of time series predictions.
Sequence Matters: Harnessing Video Models in 3D Super-Resolution
Hyun-kyu Ko (Sungkyunkwan University), Eunbyung Park (Sungkyunkwan University)
RestorationSuper ResolutionImageVideo
🎯 What it does: This paper proposes using a video super-resolution (VSR) model for three-dimensional super-resolution of low-resolution multi-view images, employing a greedy and adaptive subsequence sorting to generate 'video-like' inputs, avoiding fine-tuning of the VSR model or generating smooth trajectories.
Sequential Conditional Transport on Probabilistic Graphs for Interpretable Counterfactual Fairness
Agathe Fernandes Machado (Universite du Quebec a Montreal), Ewen Gallic (Aix Marseille University)
Explainability and InterpretabilityComputational EfficiencyGraph Neural NetworkTabular
🎯 What it does: A sequential transport method based on conditional univariate optimal transport is proposed to generate counterfactual samples that conform to causal graphs, thereby assessing individual-level fairness.
Sequential Joint Dependency Aware Human Pose Estimation with State Space Model
Hanxi Yin (University of Amsterdam), Zhixiang Chen (University of Sheffield)
Pose EstimationGraph Neural NetworkTransformerImage
🎯 What it does: This paper proposes a single-camera 2D-to-3D human pose estimation model that incorporates sequential joint dependency perception, utilizing a state space model (SSM) to integrate the sequential relationships of joints into feature learning.
Set-Valued Sensitivity Analysis of Deep Neural Networks
Xin Wang (University of Washington), Xuegang (Jeff) Ban (Southwest Jiaotong University)
Convolutional Neural NetworkImage
🎯 What it does: This paper proposes a deep neural network (DNN) sensitivity analysis framework based on set-valued mappings, studying the impact of training data perturbations on the model weight set.
SGDiff: Scene Graph Guided Diffusion Model for Image Collaborative SegCaptioning
Xu Zhang (Hunan University), Zhiyong Li (Hunan University)
SegmentationGenerationTransformerDiffusion modelContrastive LearningImageTextMultimodality
🎯 What it does: This paper proposes the task of image co-segmentation and description (SegCaptioning) and implements the generation of multiple pairs of semantically aligned captions and masks from simple prompts (such as a box) based on a Scene Graph Guided Diffusion Model (SGDiff).
SGFormer: Semantic-Geometry Fusion Transformer for Multi-modal 3D Panoptic Segmentation
Hongqi Yu (Wenzhou University), Xiaoqin Zhang (Quzhou University)
SegmentationAutonomous DrivingTransformerMultimodalityPoint Cloud
🎯 What it does: This paper proposes SGFormer, a multimodal Transformer that combines semantic and geometric information for 3D panoramic segmentation.
SGTC: Semantic-Guided Triplet Co-training for Sparsely Annotated Semi-Supervised Medical Image Segmentation
Ke Yan (Ocean University of China), Zhi Liu (Shandong University)
SegmentationConvolutional Neural NetworkContrastive LearningBiomedical DataMagnetic Resonance Imaging
🎯 What it does: This paper proposes a semantic-guided triple co-training framework SGTC, which achieves semi-supervised learning for medical image segmentation by labeling three orthogonal slices with a small number of volume samples.
Sharpening Neural Implicit Functions with Frequency Consolidation Priors
Chao Chen (Tsinghua University), Zhizhong Han (Wayne State University)
GenerationData SynthesisOptimizationPoint Cloud
🎯 What it does: Utilizing frequency-integrated priors to transform low-frequency SDF observations into high-frequency complete SDF, thereby enhancing the detail and integrity of neural implicit models.
Sharper Error Bounds in Late Fusion Multi-view Clustering with Eigenvalue Proportion Optimization
Liang Du (Shanxi University), Yuhua Qian (Shanxi University)
OptimizationImageBiomedical Data
🎯 What it does: This paper proposes a stricter analysis of the generalization error of multiple kernel K-means using the eigenvalue ratio and local Rademacher complexity, and based on this, designs a low-pass graph filter to enhance the eigenvalue ratio, thereby improving the performance of multi-view late fusion clustering.
Shield Synthesis for LTL Modulo Theories
Andoni Rodríguez (IMDEA Software Institute), Guy Katz (University of California)
OptimizationSafty and PrivacyReinforcement LearningSequential
🎯 What it does: This paper researches and implements an LTL-T shield synthesis method for systems containing numerical data, capable of intercepting and correcting dangerous behaviors of deep reinforcement learning controllers in real-time during operation.
ShotVL: Human-Centric Highlight Frame Retrieval via Language Queries
Wangyu Xue (Tsinghua University), Yaoxue Zhang (Tsinghua University)
ClassificationRetrievalTransformerSupervised Fine-TuningVision Language ModelVideoTextBenchmark
🎯 What it does: The BestShot task is proposed, which aims to accurately locate highlight frames in human videos through language queries, and a corresponding benchmark has been constructed.
SIDL: A Real-World Dataset for Restoring Smartphone Images with Dirty Lenses
Sooyoung Choi (Soongsil University), Heewon Kim (Soongsil University)
RestorationConvolutional Neural NetworkTransformerDiffusion modelImage
🎯 What it does: This paper presents a new real-world dataset - SIDL, aimed at restoring image distortions caused by stains (such as water droplets, fingerprints, dust, scratches, etc.) on smartphone cameras, and evaluates various existing restoration models on this dataset.
SIGMA: Selective Gated Mamba for Sequential Recommendation
Ziwei Liu (City University of Hong Kong), Xiangyu Zhao (City University of Hong Kong)
Recommendation SystemRecurrent Neural NetworkSequential
🎯 What it does: This paper proposes the SIGMA framework, which utilizes Partial Flip Mamba (PF-Mamba), Dense Selective Gate (DS Gate), and Feature Extraction GRU (FE-GRU) to improve sequence recommendation systems.
Sign-IDD: Iconicity Disentangled Diffusion for Sign Language Production
Shengeng Tang (Hefei University of Technology), Richang Hong (Hefei University of Technology)
GenerationPose EstimationDiffusion modelVideoText
🎯 What it does: This study proposes a sign language gesture generation framework called Sign-IDD based on diffusion models, which achieves high-quality generation of sign language gestures from text gloss by transforming 3D joint representations into 4D skeletal representations and incorporating an attribute-controllable diffusion module.