CENG0049 Deep Reinforcement Learning

Information

Course Goal / Objective

The goal of a Deep Reinforcement Learning course is to provide students with a comprehensive understanding of the principles, algorithms, and applications of reinforcement learning (RL) in combination with deep learning techniques. The course aims to equip students with the knowledge and skills necessary to design, implement, and deploy deep RL algorithms to solve complex sequential decision-making problems.

Course Content

This course covers the Introduction to Reinforcement Learning, Basics of Deep Learning, Deep Q-Networks (DQN), Q-Learning and value-based methods, Experience replay and target networks, Extensions and improvements to DQN (e.g., Double DQN, Dueling DQN), Policy Gradient Methods, Proximal Policy Optimization (PPO), Trust Region Policy Optimization (TRPO), Proximal Policy Optimization algorithms, Exploration and Exploitation, Exploration in deep reinforcement learning, Model-Based Reinforcement Learning, Advanced Topics in Deep RL, Deep RL for Robotics and Control, Deep RL in Natural Language Processing, Deep RL in Game Playing, Deep RL agents for game playing (e.g., AlphaGo, AlphaZero), Techniques for value function estimation in games, Monte Carlo methods and tree search algorithms, Ethical and Societal Considerations, Research Papers and Recent Advancements.

Course Precondition

Knowledge of basic programming, linear algebra, and probability theory.

Resources

Sewak, M. (2019). Deep reinforcement learning. Singapore: Springer Singapore.

Notes

Sutton, Richard S., and Andrew G. Barto. Reinforcement learning: An introduction. MIT press, 2018.

Course Learning Outcomes

Relation with Program Learning Outcome

Week Plan

Student Workload - ECTS