Research

Research Interests

Machine Learning, Deep Learning, Time-Series, Computer Vision, Parameter Efficient Tarnsfer Learning

Current Research Projects

Deep Learning Methods for Time-Series and Computer Vision.

Developing signal processing and machine learning techniques to improve sound classification using foundation transformer models and parameter efficient fine-tuning with histogram layers.

Multimodal AI to include text descriptions of the classifications.

Past Research Projects

Physiological Signals Analysis and Prediction with AI Algorithms.

Physics-informed neural networks for modeling cardiovascular dynamics with reduced ground truth. Implemented PINNs to incorporate known physiological constraints, achieving predictions with limited training data.

Low-Power Sensor for Human Mental Stress Diagnosis.

Applied signal processing and machine learning algorithms to identify stress in physiological data. Included microcontroller programming, schematics and PCBs design, and collection of signals through human subjects.