Register now for our CAO Open Day.
  1. Home
  2. Study
  3. Modules

Module Overview

Reinforcement Learning and Decision Making

"Reinforcement Learning and Decision Making" is a five-credit course that explores the realm of Reinforcement Learning, which is a sub-area of Machine Learning and falls under the broader umbrella of Artificial Intelligence. This course goes beyond the basics and places emphasis on computational decision-making processes that have been honed through classic and contemporary research. It covers efficient algorithms for planning for both single-agent and multi-agent scenarios and explores techniques for learning near-optimal decision making from experience.

The curriculum encompasses a wide range of topics, including Markov decision processes, stochastic games, and interactive reinforcement learning, with a special focus on the intricacies of generalization, exploration, and representation. Throughout the course, students will not only grasp the theoretical underpinnings but also gain practical expertise through assignments and a final project, preparing them for real-world applications in uncertain environments.

Module Code

CMPU 4096

ECTS Credits

10

*Curricular information is subject to change
  • Introduction to Reinforcement Learning Concepts
    • RL task formulation (Action space, State space, Environment definition)
    • Markov Decision Process and Bellman Equations
    • Defining RL environments
  • Tabular-Based Solutions
    • Dynamic Programming
    • Monte Carlo Methods
    • Temporal-Difference Learning
    • On-Policy and Off-Policy Learning
    • Q-Learning and SARSA Algorithms
  • Value Function Approximation
    • Linear Value Function Approximation
    • Non-Linear Value Function Approximation
      • Deep Q-Networks
  • Policy Gradient Methods
    • Basic: REINFORCE Algorithm
    • Advanced: Actor-Critic Algorithms
      • Proximal Policy Optimization (PPO)
      • Deep Deterministic Policy Gradient (DDPG)
  • Multi-Agent Reinforcement Learning
  • Ethics & Safety in AI
  • Practical RL Applications

 

 

 

 

 

 

 

 

 

 

 

 

 

 

This module will run for one semester, with a weekly schedule of four hours. It includes lectures, tutorials, and practical lab sessions to teach students about reinforcement learning. Lectures cover essential concepts, tutorials help clarify ideas, and lab sessions allow hands-on practice. Topics include reinforcement learning basics, prediction, control, function approximation, and policy gradients.

The course includes two projects, applying what's been learned. This module aims to give students a solid understanding of reinforcement learning for machine learning and AI.

Module Content & Assessment
Assessment Breakdown %
Formal Examination50
Other Assessment(s)50