Checking date: 08/05/2023

Course: 2023/2024

Reinforcement Learning
Master in Applied Artificial Intelligence (Plan: 475 - Estudio: 378)


Department assigned to the subject: Computer Science and Engineering Department

Type: Electives
ECTS Credits: 3.0 ECTS


Requirements (Subjects that are assumed to be known)
Machine Learning contents are recommended for the Reinforcement Learning course
Description of contents: programme
Introduction to Reinforcement Learning - Introduction to reinforcement learning - Markov Decision Processes - Policies and optimality: discounted infinite horizon - Value Functions Dynamic Programming - Problem Solving on MDP: model-free, model-based and dynamic programming methods - Policy Iteration Algorithm - Value Iteration Algorithm Direct reinforcement learning - Monte Carlo methods: and Monte Carlo with exploratory start - Model-free methods: Q-Learning - Example of execution of Q-Learning - On-policy methods vs. off-policy: SARSA - Exploration and exploitation: e-greedy and softmax Model-Based Methods - Model Learning - Dyna-Q Representation in Reinforcement Learning - Representation of the space of states, actions and Q - State space discretization: uniform and adaptive methods - Approximate methods to represent the function Q: Batch Q-Learning Generalization Through Function Approximation - Approximation through neural networks - Deep reinforcement learning Policy Search Methods - Policy Approximation - Actor-critic methods - Proximal Policy Optimization (PPO) Other Reinforcement Learning topics - Hierarchical Reinforcement Learning - Transfer of learning learned - Multi-agent Reinforcement Learning - Safe reinforcement learning - Offline Reinforcement Learning - Multi-objective Reinforcement Learning - Partially observable Reinforcement Learning Reinforcement Learning in the real world: - Applications of reinforcement learning - Reinforcement learning frameworks and software
Assessment System
  • % end-of-term-examination 30
  • % of continuous assessment (assigments, laboratory, practicals...) 70
Calendar of Continuous assessment
Basic Bibliography
  • Richard Sutton and Andrew Barto. Reinforcement Learning: an Introduction. The MIT Press.
Recursos electrónicosElectronic Resources *
(*) Access to some electronic resources may be restricted to members of the university community and require validation through Campus Global. If you try to connect from outside of the University you will need to set up a VPN

The course syllabus may change due academic events or other reasons.