Reinforcement Learning for Intelligent Decision Making

Graduate course covering reinforcement learning from core foundations to advanced RL and MARL.

Instructor: Jiadong Yu

Term: Fall

Location: HKUST(GZ)

Time: 2024/2025 Fall

Course Overview

This MPhil/PhD course systematically covers reinforcement learning from fundamental concepts to advanced methods, with an emphasis on practical implementation and research-oriented understanding.

  • Multi-armed bandits and decision-making foundations
  • Markov decision processes and value functions
  • Model-free and model-based reinforcement learning
  • Deep reinforcement learning
  • Advanced RL and multi-agent reinforcement learning

Offering

  • IOTA 5201
  • Semester: 2024/2025 Fall
  • Level: MPhil/PhD

Topics

  • Multi-armed bandits and sequential decision-making
  • Monte Carlo, temporal difference learning, Q-learning, and SARSA
  • Policy gradient and actor-critic methods
  • Deep Q-learning, PPO, TRPO, and DDPG
  • RLHF and multi-agent reinforcement learning
  • Practical coding and application-driven RL study

Teaching Focus

The course moves from RL foundations to advanced RL and MARL, helping postgraduate students connect algorithmic understanding with implementation and research applications.