Working Papers

• • [AI for Digital Pathology and Veterinary Imaging] 

    • Dual-branch Semi-supervised Adaptation of Segment Anything Model for Renal Tubule and Leukocyte Analysis, (in revision)

    • MixSCon: Manifold mix up-based supervised contrastive learning for robust histopathological pattern classification, (in revision)

    • S3DM: Semi-Supervised Signed Distance Maps for Boundary-Aware Tissue Segmentation, (in revision)

    • AnatMorph: Anatomy-Aware Deformable Image Registration for Pose Correction of Veterinary Thoracic Radiographs, (under review)

    • Decouple: Uncertainty-Weighted Event–Censor Pseudo-Supervision for Semi-Supervised Survival Learning, (under review)

    • S3D-NAS: Self-Distilled Dirichlet Distribution Neural Architecture Search in Medical Image Segmentation, (in preparation)

    • Normal Sparse Attention Pooling: A Parameter-Free Sparse Aggregation Operator for MIL Whole-Slide Image Classification, (in preparation)

    • Deep Learning-Based Digital Pathology: Feature Matching for WSI Registration Problem and Active Pseudo Label Learning for Small Data Problem, (in preparation)

• • [AI in Rehabilitation]

    • Towards Robust Skeleton-Based Rehabilitation Exercise Evaluation under Uncertain Human Pose Estimation, (in revision)

    • Feasibility and Preliminary Functional Changes Following a Gamified Hybrid Exercise Intervention in Children with Cerebral Palsy, (in revision)

    • Structure-Aware Skeleton Augmentation for Rehabilitation Exercise Recognition Across Disability Types, (in revision)

• • [AI for Autonomous Machines and Mobility] 

    • Real-Time Finite-State-Machine–Based Autonomous Lifter–Pallet Alignment, (in revision)

    • Trajectory-Aware Pedestrian Crossing and Time-to-Crosswalk Prediction with Calibration (under review)

    • Reinforcement Learning for Adaptive Control of Milling Processes Based on Cutting Force and Tool Wear, (in preparation)

• • [Robust and Adaptive Machine Learning]

    • Progressive Curriculum Learning with Adaptive Teacher Supervision for Reliable Knowledge Distillation, (under review)

    • Robust and Interpretable Anomaly Detection Using Spatiotemporal Attention for Evaluating Production Capacity, (under review)

    • Probabilistic-Robust Loss with Uncertainty-Aware Learning for Irregular Demand Patterns, (in preparation)

    • Enhanced Patch Representations for Training-Free Open-Vocabulary Semantic Segmentation, (in preparation)