Planning and Reinforcement Learning

2019-2020

Course Objective

This course has a threefold objective:
1) Understand the models and methods for sequental decision-theoretic
planning and reinforcement learning (Knowledge and understanding)
2) To understand how to model sequential decision problems using these
models. (Knowledge and understanding)
3) To gain a thorough understaning of the algorithms, and gain hands-on
experience, by performing computational experiments with planning and
reinforcement learning algorithms. (Apply knowledge and understanding)
(Make judgements)

Course Content

This course will cover basic as well as advanced concepts of
Reinforcement Learning and Decision-Theoretic Planning (also known as
planning under uncertainty).

This course is an introduction to the basic concepts that underlie the
design of autonomous agents in modern artificial intelligence. It
addresses fundamental challenges such as how such agents can maximise
their utility by planning sequences of actions, and learn by trial and
error. In order to do so, these agents must cope with uncertainty about
their environment, and may have to cooperate and/or compete with other
agents.

Specific topics covered include: Introduction to Autonomous Agents and
sequential decision problems, Multi-armed Bandits, dynamic programming
(planning),
Monte Carlo methods (planning and learning), temporal difference
methods, Model-based reinforcement learning, (RL) Learning Theory,
Partial observability,
Cooperative Multi-agent planning and learning, Self-interested
Multi-agent learning,
Multi-objective planning and learning, and least but not least, Deep
Reinforcement Learning.

Teaching Methods

Oral lectures and compulsory programming assignment (in teams of 3 or
4).

Method of Assessment

Written exam and programming assignment (weighted average). To pass the
course as a whole, you must pass both the exam and the programming
assignment.

Entry Requirements

Programming skills are necessary to do the practical assignment.

Literature

Sutton and Barto – Introduction to Reinforcement Learning (available
both online, and as printed book)
The slides of the course
Additional papers (TBA during the course)

Target Audience

MSc Business Analytics
MSc Computer Science
MSc Artificial Intelligence

General Information

Course Code XM_0055
Credits 6 EC
Period P2
Course Level 400
Language of Tuition English
Faculty Faculty of Science
Course Coordinator dr. M. Hoogendoorn
Examiner dr. M. Hoogendoorn
Teaching Staff

Practical Information

You need to register for this course yourself

Last-minute registration is available for this course.

Teaching Methods Seminar, Lecture
Target audiences

This course is also available as: