Research Works

Deep Learning Based Early Prediction of Depression and Suicidal Risk in Bangla Social Media Posts

Presented at STI 2025. This research develops deep learning models to detect early signs of depression and suicidal risk from Bangla social media content. It integrates text preprocessing, sentiment analysis, and neural architectures for high-accuracy risk prediction.

Transformer-Based Hybrid Framework for Early Detection of Depression and Suicidal Risk

Submitted to QPAIN 2026. This framework combines Transformer and BiLSTM models to enhance early mental health detection, using multimodal features and explainable AI for identifying depressive and suicidal patterns.

Retrieval-Grounded and Novelty-Aware Framework for Research Idea Generation Using LLMs

Submitted to QPAIN 2026. This research leverages Large Language Models (LLMs) to automatically generate, evaluate, and assess novelty of research and project ideas, assisting students and researchers in discovering innovative directions.

LLM–Driven Reinforcement Learning Framework for Employee Performance Optimization

Submitted to QPAIN 2026. This work integrates LLMs with reinforcement learning to optimize employee performance, combining human feedback and automated evaluation metrics for organizational efficiency.

Enhancing Reinforcement Learning Alignment via LLM-Based Interpretation of Human Feedback

Submitted to QPAIN 2026. This research improves reinforcement learning alignment by interpreting human feedback with LLMs, aiming for better reward shaping and policy optimization in AI systems.