A PyTorch reinforcement learning library centered around reproducible and generalizable algorithm implementations.

A PyTorch library to easily facilitate knowledge distillation for custom deep learning models

Neural Correlates for Reinforcement Learning

A review of the connections between neuroscience and reinforcement learning.


An exploratory and experimental research into the applications of modern deep learning breakthroughs in traditional time-series analysis approaches.


Alzheimer's Dementia Recognition from Spontaneous Speech. The aim is unbiased early detection of cognitive decline from multi-modal data.

SNNs to Validate Experimental Results

Building computational spiking neural network models to model neural functioning and thereby validate experimental results.

A Python Library for Robustness Monitoring and Adversarial Debugging of NLP models.


DAFL(Data free learning) is an unsupervised Knowledge Distillation technique. We are trying to apply IKD on this technique for smaller datasets as of now like MNIST/CIFAR.

A python library consisting of pipelines for visual analysis of different sports using Computer Vision and Deep Learning.

Sponsored Projects

Flatbuffer/Protobuffer Integration for VowpalWabbit

Principal Investigator: Sharad Chitlangia
Grant Amount: 10000$
Funding Agency: Microsoft Research
Short description of proposal: Addition of new serialization protocols for input examples, model files in VowpalWabbit for extremely fast and scalable Machine Learning.

Past Projects

Stock market Modelling using Reinforcement Learning

This project in collaboration with ISI Kolkata aims to apply deep Reinforcement learning to financial trading markets in order to develop a Markov decision process which will suitably capture any general financial market.

Implementation of Neural Voice Cloning with Few Samples paper by Baidu Research

Drone performing imitation learning on IDSIA dataset. Project funded by EEE department and Academic Graduate Studies & Research Division

Word2Brain2Image: Visual Reconstruction from Spoken Word Representations

Collected EEG data of subjects listening to spoken object words. We attempted to use deep generative models to generate images of objects using only this EEG data, in order to study internal representations in visual association areas (BA 18,19) of the brain.

Research Project in unofficial collaboration with TCS Research. Applying State of the art models for pneumonia detection on RSNA pneumonia detection dataset. Tested InceptionNet-v3, DenseNet121 and explored Mask RCNN applicability for the dataset. Got 83.8% and 77.9% classification accuracy respectively.

Emotion Recognition From EEG Signals

A project with the aim to use deep convolutional networks on EEG signals in order to predict valence and arousal values from them.

Dynamic Gesture Recognition for controlling portable devices

Using detected dynamic gestures such as swiping left/right, closing fist, etc. to control portable devices ( multitasking / controlling video playback, etc. )