Undergrad Research Project - Reinforcement Learning

Fall 2016

Student
Mario Srouji
Advisor
Ruslan Salakhutdinov
Project description

The project consists of learning the fundamentals of Reinforcement Learning, which is a field of Machine Learning, in order to solve a variety of problems. RL algorithms can be used to solve control problems (robots performing activities autonomously), and eventually lead up to implementing much more complicated algorithms, that can learn an infinite state space, with an infinite action space. Much of the project requires independent reading, in order to understand a lot of the concepts and algorithms. In addition, I have been working in open AI's gyms, which is a collection of simulation environments that allows me to test my algorithms against realistic problems. I meet weekly to bi-weekly with my Professor to go over the progress, and discuss the results. The goals of this project is to continue to understand deeper RL fundamentals, and implement more complex algorithms to solve more complex problems.

Return to project list