Neelesh Kumar

I am a fifth year PhD Candidate at Computational Brain Lab, Rutgers University, advised by Prof. Konstantinos Michmizos. My main research interest is in developing machine learning algorithms for EEG, with applications in Neurorehabilitation robots. I also work on developing neuromorphic algorithms for supervised and reinforcement learning.

Email  /  CV  /  Google Scholar

A brain-inspired 3D-Convolutional Neural Network that can predict the fundamental components of movements from EEG using features that correspond with the underlying neurophysiology.

A biologically plausible gradient-based learning rule for SNNs that is as effective as backprop, and can also be deployed on Intel's Loihi for energy-efficient learning.

A CNN that can accurately predict movement intent and reaction time from the EEG before the movement actually takes place.

A CNN to objectively assess in real-time how engaged subjects are in performing rehabilitation tasks.

SNNs for popular DRL algorithms that perform as well as their ANN counterparts while consuming 140 times less energy per inference when deployed on Intel's Loihi

A hybrid SNN-ANN reinforcement learning algorithm that navigates a mobile robot effectively with high energy-efficiency