Selected work

Representative Papers

Uncertainty-Aware and Explainable Ensemble Deep Learning Framework for Multi-Class Skin Lesion Classification

Aug 11, 2026

This study addresses the challenges of multi-class skin lesion classification in dermoscopic images—namely, high intra-class variability, inter-class similarity, class imbalance, and insufficient model interpretability—by proposing a deep ensemble framework that integrates MaxViT-Tiny with multiple convolutional neural networks (ConvNeXt-Tiny and EfficientNetV2-B0). To the best of our knowledge, this is the first work to apply MaxViT-Tiny to skin lesion classification. The framework incorporates Monte Carlo Dropout to quantify prediction uncertainty and employs Grad-CAM++ for visual interpretability. Evaluated on the HAM10000 dataset, the model achieves 96% accuracy, 99% ROC-AUC, and 95% macro-averaged precision, recall, and F1-score after uncertainty-based filtering, substantially enhancing diagnostic reliability and transparency.

0 citationsRead paper

FedLLM: A Privacy-Preserving Federated Large Language Model for Explainable Traffic Flow Prediction

Apr 17, 2026

This work addresses key limitations in existing traffic flow prediction methods—namely their reliance on centralized data, limited interpretability, and vulnerability to data heterogeneity and privacy concerns. To overcome these challenges, the authors propose FedLLM, a novel framework that integrates federated learning with domain-adapted large language models. By leveraging structured traffic prompts, lightweight LoRA-based parameter exchange, and a composite selection-scoring mechanism, FedLLM enables high-performance multi-step short-term forecasting while preserving data locality and privacy. The approach effectively mitigates non-IID data distributions, achieves prediction accuracy surpassing centralized baselines, and produces interpretable, structurally coherent reasoning outputs with strong cross-regional generalization capabilities.

0 citationsRead paper

A Hierarchical Importance-Guided Multi-objective Evolutionary Framework for Deep Neural Network Pruning

Apr 01, 2026

This work addresses the high-dimensional, large-scale, and highly non-convex optimization challenges inherent in pruning over-parameterized deep neural networks by proposing a layer-wise importance-guided multi-objective evolutionary framework. The method formulates pruning as a large-scale multi-objective optimization problem and employs a two-stage cooperative mechanism combining coarse-grained continuous search with fine-grained binary optimization. Leveraging importance-aware sampling, adaptive mutation, and Pareto-front refinement, it efficiently explores the exponential decision space to identify Pareto-optimal solutions balancing model accuracy and compactness. Experiments on CIFAR-10 and CIFAR-100 demonstrate that the approach achieves parameter compression rates of up to 51.9% for ResNet-56 and 38.9% for ResNet-110 with negligible accuracy loss, significantly outperforming existing evolutionary pruning methods.

0 citationsRead paper

Med-SegLens: Latent-Level Model Diffing for Interpretable Medical Image Segmentation

Feb 11, 2026

This work addresses the limited interpretability of medical image segmentation models, which hinders error diagnosis and robustness under data distribution shifts. The authors propose the first latent-level differential framework tailored for medical image segmentation, leveraging sparse autoencoders to extract interpretable latent variables from the internal representations of SegFormer and U-Net. By systematically analyzing representation discrepancies across architectures and datasets, they uncover how shared and population-specific latent factors influence model performance. Building on these insights, they enable causal interventions without retraining, successfully restoring segmentation accuracy in 70% of failure cases—boosting the Dice score from 39.4% to 74.2%—and substantially improving cross-dataset generalization.

0 citationsRead paper

Real-Time Drivers' Drowsiness Detection and Analysis through Deep Learning

Nov 15, 2025

To address road safety risks posed by driver fatigue, this paper proposes a non-intrusive, low-cost, real-time detection method leveraging a deep convolutional neural network (DCNN) and OpenCV. The system captures facial video streams via an in-vehicle camera, employs a pre-trained model combined with facial landmark detection to accurately quantify fatigue indicators—specifically PERCLOS (Percentage of Eyelid Closure) and yawning—and triggers multi-level alerts accordingly. Evaluated on the NTHU-DDD and Yawn-Eye-Dataset benchmarks, the method achieves 99.6% and 97.0% classification accuracy for fatigue states, respectively, outperforming existing lightweight approaches. Our key contributions include: (1) a fully end-to-end, deployable real-time framework; (2) a balanced design achieving high accuracy with low computational overhead; and (3) successful integration into an intelligent in-vehicle platform, with robustness and sub-second response latency empirically validated under real-world driving conditions.

0 citationsRead paper

Recent publications

Latest Papers

Uncertainty-Aware and Explainable Ensemble Deep Learning Framework for Multi-Class Skin Lesion Classification

Aug 11, 2026

This study addresses the challenges of multi-class skin lesion classification in dermoscopic images—namely, high intra-class variability, inter-class similarity, class imbalance, and insufficient model interpretability—by proposing a deep ensemble framework that integrates MaxViT-Tiny with multiple convolutional neural networks (ConvNeXt-Tiny and EfficientNetV2-B0). To the best of our knowledge, this is the first work to apply MaxViT-Tiny to skin lesion classification. The framework incorporates Monte Carlo Dropout to quantify prediction uncertainty and employs Grad-CAM++ for visual interpretability. Evaluated on the HAM10000 dataset, the model achieves 96% accuracy, 99% ROC-AUC, and 95% macro-averaged precision, recall, and F1-score after uncertainty-based filtering, substantially enhancing diagnostic reliability and transparency.

0 citationsRead paper

FedLLM: A Privacy-Preserving Federated Large Language Model for Explainable Traffic Flow Prediction

Apr 17, 2026

This work addresses key limitations in existing traffic flow prediction methods—namely their reliance on centralized data, limited interpretability, and vulnerability to data heterogeneity and privacy concerns. To overcome these challenges, the authors propose FedLLM, a novel framework that integrates federated learning with domain-adapted large language models. By leveraging structured traffic prompts, lightweight LoRA-based parameter exchange, and a composite selection-scoring mechanism, FedLLM enables high-performance multi-step short-term forecasting while preserving data locality and privacy. The approach effectively mitigates non-IID data distributions, achieves prediction accuracy surpassing centralized baselines, and produces interpretable, structurally coherent reasoning outputs with strong cross-regional generalization capabilities.

0 citationsRead paper

A Hierarchical Importance-Guided Multi-objective Evolutionary Framework for Deep Neural Network Pruning

Apr 01, 2026

This work addresses the high-dimensional, large-scale, and highly non-convex optimization challenges inherent in pruning over-parameterized deep neural networks by proposing a layer-wise importance-guided multi-objective evolutionary framework. The method formulates pruning as a large-scale multi-objective optimization problem and employs a two-stage cooperative mechanism combining coarse-grained continuous search with fine-grained binary optimization. Leveraging importance-aware sampling, adaptive mutation, and Pareto-front refinement, it efficiently explores the exponential decision space to identify Pareto-optimal solutions balancing model accuracy and compactness. Experiments on CIFAR-10 and CIFAR-100 demonstrate that the approach achieves parameter compression rates of up to 51.9% for ResNet-56 and 38.9% for ResNet-110 with negligible accuracy loss, significantly outperforming existing evolutionary pruning methods.

0 citationsRead paper

Med-SegLens: Latent-Level Model Diffing for Interpretable Medical Image Segmentation

Feb 11, 2026

This work addresses the limited interpretability of medical image segmentation models, which hinders error diagnosis and robustness under data distribution shifts. The authors propose the first latent-level differential framework tailored for medical image segmentation, leveraging sparse autoencoders to extract interpretable latent variables from the internal representations of SegFormer and U-Net. By systematically analyzing representation discrepancies across architectures and datasets, they uncover how shared and population-specific latent factors influence model performance. Building on these insights, they enable causal interventions without retraining, successfully restoring segmentation accuracy in 70% of failure cases—boosting the Dice score from 39.4% to 74.2%—and substantially improving cross-dataset generalization.

0 citationsRead paper

Real-Time Drivers' Drowsiness Detection and Analysis through Deep Learning

Nov 15, 2025

To address road safety risks posed by driver fatigue, this paper proposes a non-intrusive, low-cost, real-time detection method leveraging a deep convolutional neural network (DCNN) and OpenCV. The system captures facial video streams via an in-vehicle camera, employs a pre-trained model combined with facial landmark detection to accurately quantify fatigue indicators—specifically PERCLOS (Percentage of Eyelid Closure) and yawning—and triggers multi-level alerts accordingly. Evaluated on the NTHU-DDD and Yawn-Eye-Dataset benchmarks, the method achieves 99.6% and 97.0% classification accuracy for fatigue states, respectively, outperforming existing lightweight approaches. Our key contributions include: (1) a fully end-to-end, deployable real-time framework; (2) a balanced design achieving high accuracy with low computational overhead; and (3) successful integration into an intelligent in-vehicle platform, with robustness and sub-second response latency empirically validated under real-world driving conditions.

0 citationsRead paper