Research Intern - EEG Signal Processing for Speech Imagery Decoding (IIT Bhubaneswar)
Conducted EEG preprocessing and supervised learning workflows for imagined and spoken speech decoding using Python-based signal processing and machine learning. Performed electrode-importance analysis on 64-channel EEG signals to interpret neural contribution patterns using SHAP-based methods. The work involved preparing model-ready inputs from electrophysiology recordings and evaluating classification performance for speech imagery tasks. • EEG data used for supervised classification of imagined vs spoken speech • SHAP-based interpretability to assess electrode contribution • Signal preprocessing and feature/representation learning pipeline development • Model evaluation using classification metrics across decoding scenarios