Research Blogs News Tags Projects

Research

Select a Tag

Published on
June 5, 2025
AetherVision-Bench: An Open-Vocabulary RGB-Infrared Benchmark for Multi-Angle Segmentation across Aerial and Ground Perspectives
open-vocabulary-semantic-segmentation embodied-AI-systems robustness-benchmark cross-domain-generalization aerial-perspectives multi-angle-segmentation autonomous-navigation zero-shot-transfer-models ground-perspectives
• Robert Bosch Centre for Cyber Physical Systems, Indian Institute of Science, Bengaluru, India• Kotak IISc AI-ML Centre, Indian Institute of Science, Bengaluru, India• Department of Aerospace Engineering, Indian Institute of Science, Bengaluru, India
Published on
June 5, 2025
OV-COAST: Cost Aggregation with Optimal Transport for Open-Vocabulary Semantic Segmentation
open-vocabulary-semantic-segmentation out-of-domain-generalization CLIP Sinkhorn-distance cost-aggregation MESS-benchmark optimal-transport cost-volume CAT-Seg-model ViT-B-backbone
• Kotak IISc AI-ML Centre, Indian Institute of Science, Bengaluru, India• Robert Bosch Centre for Cyber Physical Systems, Indian Institute of Science, Bengaluru, India• Department of Aerospace Engineering, Indian Institute of Science, Bengaluru, India
Published on
June 5, 2025
ScoreRAG: A Retrieval-Augmented Generation Framework with Consistency-Relevance Scoring and Structured Summarization for News Generation
journalistic-standards vector-database large-language-models hallucinations graded-summaries structured-summarization automated-news-generation Natural-Language-Processing consistency-relevance-evaluation factual-inconsistencies domain-specific-expertise retrieval-augmented-generation
• Department of Mathematical Sciences, National Chengchi University, Taiwan
Published on
June 5, 2025
Learning-at-Criticality in Large Language Models for Quantum Field Theory and Beyond
arithmetic-reasoning large-language-models scale-free-exploration learning-at-criticality concept-network-model quantum-field-theory critical-thinking-pattern reinforcement-learning information-scarcity second-order-phase-transition symbolic-Matsubara-sums generalization
Published on
June 5, 2025
AdaDecode: Accelerating LLM Decoding with Adaptive Layer Parallelism
speculative-decoding decoding-efficiency hardware-utilization large-language-models parallel-processing token-generation autoregressive-decoding decoding-latency output-consistency layer-skipping long-content-generation
• Department of Physics, University of Science and Technology• Department of Chemistry, National Taiwan University• Institute of Materials Science, Tsinghua University• Department of Electrical Engineering, Stanford University
Published on
June 5, 2025
Advancements in Artificial Intelligence Applications for Cardiovascular Disease Research
deep-learning-architectures generative-adversarial-networks validation-protocols precision-diagnostics automated-analysis convolutional-neural-networks cardiovascular-medicine multimodal-data diagnostic-accuracy adaptive-algorithms personalized-cardiovascular-care ultrasound workflow-efficiency magnetic-resonance-imaging artificial-intelligence computed-tomography electrocardiography
Published on
June 5, 2025
Comprehensive Attribute Encoding and Dynamic LSTM HyperModels for Outcome Oriented Predictive Business Process Monitoring
simultaneous-event-modeling hierarchical-encoding explainable-process-intelligence heterogeneous-datasets multidimensional-embeddings adaptive-representations multi-level-attributes Predictive-Business-Process-Monitoring dynamic-LSTM-HyperModels class-imbalance
• Department of Computer Science, University of XYZ• Department of Information Technology, University of ABC• Department of Computer Engineering, University of DEF
Published on
June 5, 2025
Robust Preference Optimization via Dynamic Target Margins
preference-optimization-algorithm high-confidence-pairs data-quality dynamic-target-margin Large-Language-Models training-efficiency noise-reduction Direct-Preference-Optimization instance-specific-margin-calibration
Published on
June 5, 2025
DSSAU-Net:U-Shaped Hybrid Network for Pubic Symphysis and Fetal Head Segmentation
pubic-symphysis-segmentation Head-Symphysis-Distance MICCAI-IUGC-2024-competition sparse-self-attention-network Dual-Sparse-Selection-Attention U-shaped-encoder-decoder-network DSSAU-Net Intrapartum-Ultrasound-Grand-Challenge fetal-head-segmentation multiscale-feature-fusion ultrasound-assisted-diagnosis invasive-vaginal-examinations computational-efficiency Angle-of-Progression childbirth-process
• Department of Physics, University of Science and Technology, China
Published on
June 5, 2025
PRJ: Perception-Retrieval-Judgement for Generated Images
content-moderation AI-generated-visual-content contextual-toxicity-risk-matrix toxicity-detection structured-reasoning-process image-safety-systems generative-AI digital-media-regulation
Published on
June 5, 2025
How PARTs assemble into wholes: Learning the relative composition of images
EEG-signals natural-images object-detection grid-based-approaches audio off-grid-patches medical-imaging continuous-relative-transformations representation-learning self-supervised-learning spatial-aware-pretext-tasks universal-self-supervised-pretraining time-series-prediction video
• Apple• Unknown
Published on
June 5, 2025
Efficient Data Selection for Domain Adaptation of ASR Using Pseudo-Labels and Multi-Stage Filtering
named-entity-recognition pseudo-labels word-error-rate Whisper-model domain-adaptation call-center-data data-selection-pipelines training-segments character-error-rate ASR-models Zipformer-model
• Department of Computer Science, University of XYZ• Department of Electrical Engineering, University of ABC• Institute of Advanced Studies, University of DEF• Department of Mechanical Engineering, University of GHI
Published on
June 5, 2025
BiXFormer: A Robust Framework for Maximizing Modality Effectiveness in Multi-Modal Semantic Segmentation
mask-level-classification mIoU-improvements Unified-Modality-Matching Cross-Modality-Alignment Complementary-Matching semantic-segmentation RGB-and-non-RGB-modalities Modality-Agnostic-Matching multi-modal-data
• Unknown
Published on
June 5, 2025
Out-of-Distribution Graph Models Merging
heterogeneous-GNN-backbones masking-mechanism graph-generation-strategy generalized-model-construction MoE-module out-of-distribution-graph-models model-generalization mixture-distribution domain-invariant-knowledge
• School of Computing and Information Technology, Great Bay University• Department of Computer Science, University of California, Los Angeles• Department of Computer and Information Science, University of Macau• School of Software Engineering, Huazhong University of Science and Technology
Published on
June 5, 2025
Reason from Future: Reverse Thought Chain Enhances LLM Reasoning
reasoning-paradigms error-accumulation empirical-evaluations goal-oriented-constraints logical-relationships complex-tasks local-optimum-reasoning reverse-reasoning-mechanism bottom-up-reasoning accuracy Reason-from-Future Chain-of-Thought global-perspective bidirectional-reasoning top-down-planning small-language-models Tree-of-Thought
• College of Computer Science and Technology, Zhejiang University, Hangzhou, China• Alibaba Group, Hangzhou, China• State Key Laboratory of Transvascular Implantation Devices and TIDRI, Hangzhou, 310009, China
Published on
June 5, 2025
Accelerating SfM-based Pose Estimation with Dominating Set
dominating-set augmented-reality robotics Structure-from-Motion real-time-applications pose-estimation graph-theory OnePose-dataset processing-speed virtual-reality 3D-pose-estimation
• Indian Institute of Science, Bangalore, Karnataka, India, 560012
Published on
June 5, 2025
ROSA: Addressing text understanding challenges in photographs via ROtated SAmpling
Visual-Question-Answering visually-impaired ROtated-SAmpling text-rich-images decoding-strategy VQA-performance text-recognition photographs
• FAMAF, Universidad Nacional de Córdoba, CONICET, Argentina
Published on
June 5, 2025
Intersectional Bias in Pre-Trained Image Recognition Models
sensitive-variables race visualization facial-images gender ImageNet-classifiers Deep-Learning activations age topographic-maps pre-trained-models linear-classifier-probes encoded-biases
• Unknown
Published on
June 5, 2025
An Improved Grey Wolf Optimizer Inspired by Advanced Cooperative Predation for UAV Shortest Path Planning
obstacle-laden-environments Lens-Opposition-based-Learning-Strategy shortest-flight-path-planning local-optima Improved-Grey-Wolf-Optimizer Advanced-Cooperative-Predation Unmanned-Aerial-Vehicles optimization-performance heuristic-based-methods
Published on
June 5, 2025
Zero-Shot Temporal Interaction Localization for Egocentric Videos
zero-shot-temporal-action-localization closed-loop-feedback self-adaptive-sampling-strategy 3D-hand-velocities human-object-interaction vision-language-models temporal-action-localization egocentric-videos
• IRMV Lab, the Department of Automation, Shanghai Jiao Tong University
Published on
June 5, 2025
INP-Former++: Advancing Universal Anomaly Detection via Intrinsic Normal Prototypes and Residual Learning
normal-references industrial-inspection Intrinsic-Normal-Prototypes reconstruction-errors medical-diagnosis anomaly-detection INP-guided-Decoder INP-Extractor Soft-Mining-Loss residual-learning INP-Coherence-Loss zero-shot-capability local-variations detection-accuracy state-of-the-art-performance
• State Key Laboratory of Precision Measurement Technology and Instruments, Department of Precision Instrument, Tsinghua University, Beijing, China• Hunan University School of Robotics, Changsha, China• Institute of Automation, Chinese Academy of Sciences, Beijing, China• State Key Laboratory of Intelligent Manufacturing Equipment and Technology, Huazhong University of Science and Technology, Wuhan, China
Published on
June 5, 2025
Trustworthy Medical Question Answering: An Evaluation-Centric Survey
Robustness trustworthiness retrieval-augmented-grounding Factuality Fairness dimensions-of-trustworthy-AI benchmarks Explainability safety-alignment future-research-directions evaluation-guided-techniques open-challenges patient-outcomes clinical-decision-making Safety large-language-models healthcare-question-answering-systems Calibration adversarial-fine-tuning
• School of Artificial Intelligence, Shandong University, China• Department of Computer Science, University of Western Ontario, Canada• Faculty of Computer Science, Dalhousie University, Canada
Published on
June 5, 2025
MambaNeXt-YOLO: A Hybrid State Space Model for Real-time Object Detection
MambaNeXt-YOLO real-time-object-detection linear-state-space-models Multi-branch-Asymmetric-Fusion-Pyramid-Network MambaNeXt-Block computer-vision Transformer-based-architectures computational-complexity edge-focused-efficiency YOLO-series-models
• Equal contribution.
Published on
June 5, 2025
EmoArt: A Multidimensional Dataset for Emotion-Aware Artistic Generation
creative-design art-therapy artistic-images diffusion-models computational-art emotion-annotated-art-datasets affective-computing emotion-driven-image-synthesis text-to-image-generation multimodal-learning
• Jilin University, Changchun, China
Published on
June 5, 2025
YOND: Practical Blind Raw Image Denoising Free from Camera-Specific Data Dependency
camera-specific-data-dependency SNR-guided-denoiser coarse-to-fine-noise-estimation learning-based-methods expectation-matched-variance-stabilizing-transform adaptive-adjustments manual-fine-tuning blind-raw-image-denoising synthetic-data
• School of Computer Science and Technology, Beijing Institute of Technology, Beijing, 100081, China• School of Artificial Intelligence, Beijing Normal University, Beijing, 100875, China
Published on
June 5, 2025
Images are Worth Variable Length of Representations
fixed-length-sequence semantic-extraction autoencoder-based-tokenization vision-encoders reconstruction-quality query-conditioned-tokenization variable-number-of-visual-tokens dynamic-vision-encoder
• University of California, San Diego• University of California, Berkeley• University of Washington
Published on
June 5, 2025
Spatial Understanding from Videos: Structured Prompts Meet Simulation Data
ScanForgeQA spatial-uncertainty 3D-spatial-reasoning 3D-simulation-scenes structured-prompting-strategy data-scarcity pre-trained-vision-language-models robotic-navigation visual-spatial-understanding embodied-interaction fine-tuning question-answering-dataset
• Harbin Institute of Technology (Shenzhen)• Shandong Jianzhu University
Published on
June 5, 2025
RewardAnything: Generalizable Principle-Following Reward Models
Large-Language-Model-optimization real-world-needs benchmark-for-reward-models natural-language-specifications Reward-Models task-specific-preference-data RABench preference-datasets generalization RLHF-methods generalizable-reward-models
• Peking University• WeChat AI• William & Mary• Westlake University
Published on
June 5, 2025
FingerVeinSyn-5M: A Million-Scale Dataset and Benchmark for Finger Vein Recognition
synthetic-generator intra-class-variations finger-vein-recognition performance-gain benchmarking finger-vein-dataset pretrained-models large-scale-public-datasets deep-learning
• School of Cyber Science and Engineering, Southeast University, Nanjing, China
Published on
June 5, 2025
Robustness of Prompting: Enhancing Robustness of Large Language Models Against Prompting Attacks
Error-Correction Guidance adversarial-examples model-accuracy input-perturbations Large-Language-Models robustness prompting-strategy
• Anhui University• Hefei University• University of Science and Technology of China