Publications

Physiologically Grounded Driver Behavior Classification: SHAP-Driven Elite Feature Selection and Hybrid Gradient Boosting for Multimodal Physiological Signals

Published in arXiv Preprint, 2026

This study introduces an interpretable ensemble framework using SHAP-based “Elite” feature selection and hybrid gradient boosting (XGBoost + LightGBM) to classify driving behaviors from multimodal physiological signals (EEG, EMG, GSR).

Download Paper

QiVC-Net: Quantum-Inspired Variational Convolutional Network, with Application to Biosignal Classification

Published in arXiv Preprint, 2025

QiVC-Net introduces the Quantum-inspired Variational Convolution (QiVC) framework, a novel learning paradigm that integrates probabilistic inference, variational optimization, and quantum-inspired transformations into convolutional neural networks.

Download Paper

Adaptive Real-Time Multi-Loss Function Optimization Using Dynamic Memory Fusion Framework: A Case Study on Breast Cancer Segmentation

Published in Biomedical Signal Processing and Control, 2025

In this research, we propose a dynamic memory fusion framework for adaptive multi-loss function penalizing in real-time, which dynamically adjusts the weighting of loss functions based on historical loss values and incorporates an auxiliary loss function and class-balanced dice loss to improve segmentation performance. Download Manuscript

Download Paper

Probabilistic Deep Learning and Transfer Learning for Robust Cryptocurrency Price Prediction

Published in Expert Systems with Applications, 2024

In this research, we propose a probabilistic gated recurrent unit (P-GRU) model for accurate Bitcoin price prediction, outperforming traditional deep learning models in capturing market volatility and uncertainty

Download Paper

DeepFaceAR: Deep Face Recognition and Displaying Personal Information via Augmented Reality

Published in International Conference on Machine Vision and Image Processing (MVIP), 2020

In this research, we propose a system that combines Deep Neural Networks and Augmented Reality to recognize faces and display information about individuals. Download Manuscript

Download Paper

Supervised Contrastive Representation Learning and Single-Stage Mixture Density Fusion for Uncertainty-Aware Classification of Driving Behavior

Published in Under Preparation, 2025

We propose a multimodal physiological modeling framework combining Supervised Contrastive Representation Learning and a unified Mixture Density Fusion module to achieve reliable, uncertainty-aware driving behavior classification.

Efficient Multi-Scale Brain Tumor Segmentation: Integrating Adaptive Distributed Attention and Weighted Inception Module

Published in Under Preparation, 2025

In this research, we propose a novel approach to enhancing the U-Net architecture for brain tumor segmentation using BRATS dataset. By integrating Weighted Inception Module and Adaptive Distributed Attention Mechanism, our proposed Dynamic U-Net achieves superior segmentation accuracy and computational efficiency compared to traditional U-Net variants.