Research

Sparse Reward Propagation for Deep Reinforcement Learning

Extracted a potential-based reward shaping function automatically from the MDP using a GCN and augumented it on the sprase reward system to propagate rewards...

Paraphrase Generator with T5 transformer

A Paraphrase-Generator built using transformers which takes an English sentence as an input and always produces a set of unique paraphrased sentences based o...

AWANTA: Adaptive Software-Defined Wide Area Network framework for Telehealth Access

This project provides an emulator designed to simulate a small, fully connected mesh network with n nodes. The primary purpose of this emulator is to demonst...

Capstone Project: MARL for Cooperative and Independent Behaviors on Highways and Ring Roads

Analyzed and studied novel coordinated MARL algorithms to avoid collisons on highways and dissipating traffic congestion waves in ring road systems. Develope...