JAWAHARLAL NEHRU TECHNOLOGICAL UNIVERSITY HYDERABAD

Course Objectives:

Learn how to define RL tasks and the core principals behind the RL, including policies, value functions. Understand and work with tabular methods to solve classical control problems. Recognize current, advanced techniques and applications in RL

Course Outcomes

Upon the Successful Completion of the Course, the Students would be able to:

Explore imitation learning tasks and solutions Learn how to define RL tasks and the core principals behind the RL, including policies, value Functions Understand and work with tabular methods to solve classical control problems.

UNIT - I

Reinforcement Learning Primitives: Introduction, Basics of RL, Defining RL Framework, Probability Basics: Probability Axioms, Random Variables, Probability Mass Function, Probability Density Function, Cumulative Distribution Function and Expectation. Introduction to Agents, Intelligent Agents – Problem Solving – Searching, Logical Agents.

UNIT - II

Markov Decision Process and Dynamic Programming: Markov Property, Markov Chains, Markov Reward Process (MRP), Bellman Equations for MRP, Dynamic Programming: Polices (Evaluation, Improvement, Iteration, Value Iteration), Asynchronous Dynamic Programming, Generalized Policy Iteration.

UNIT - III

Monte Carlo Methods and Temporal Difference Learning: Monte Carlo- Prediction, Estimation of Action Values, Control and Control without Exploring Starts, Off-Policy Control, Temporal Difference Prediction:TD(0), SARSA: On-Policy TD control, Q-Learning: Off-Policy TD control, Games

UNIT - IV

Deep Reinforcement Learning: Deep Q-Networks, Double Deep-Q Networks (DQN, DDQN, Dueling DQN, Prioritized Experience Replay), Introduction to Policy-based Methods, Vanilla Policy Gradient, REINFORCE Algorithm and Stochastic Policy Search, Asynchronous Actor Critic and Asynchronous Advantage Actor Critic (A2C, A3C), Advanced Policy Gradient (PPO, TRPO, DDPG).

UNIT - V

Multi Agent in RL: Multi-Agent Learning, Meta-learning, Partially Observable Markov Decision Process, Ethics in RL, Applying RL for Real-World Problems

Text Books:

Richard S. Sutton and Andrew G. Barto, "Reinforcement learning: An Introduction", Second Edition, MIT Press, 2019. Russell, Stuart J., and Peter Norvig. "Artificial intelligence: a modern approach.", Pearson Education Limited, 2016. Michael Wooldridge, “An Introduction to Multi Agent Systems”, John Wiley, 2002.

M. Tech. (Data Science)