This class will be given during the second semester on Wednesday afternoon. It starts at 1:45pm. The first class takes place on the 3rd of February. Given the COVID-19 crisis, the class will be given on-line using Webex. Access the class by clicking on the following link: https://uliege.webex.com/meet/dernst
The teaching assistants for the class are Samy Aittahar and Bardhyl Miftari. You should contact them using the following email address: eb.eg1614384051eilu@16143840513008o1614384051fni1614384051 .
Lesson 1. 03/02/2021. Speaker: Damien Ernst. Introduction to Reinforcement Learning (RL). Understand how to build a RL agent for non-adversarial environment with discrete state/action spaces. Podcast lesson 1
Lesson 2. 10/02/2021. Speaker: Damien Ernst. The Q-learning algorithm (see Slides Lesson 1). Proof related to the upper bound on the suboptimality of \mu_T^* Podcast Lesson 2
Lesson 4. 24/02/2021. Speaker: Damien Ernst. Advanced algorithms for learning Q-functions.
Lesson 5. 03/03/2021. Speaker: Damien Ernst. Discussion assignments, project and presentation Research paper 2.
Lesson 6. 10/03/2021. Speaker: Adrien Bolland. Gradient-based techniques for reinforcement learning in continuous domains.
Lesson 7. 17/03/2021. Speaker: Adrien Bolland. Gradient-based techniques for reinforcement learning in continuous domains.
Lesson 8. 24/03/2021. Speaker: Thibaut Théate. Distributed reinforcement learning.
Lesson 10. 21/04/2021. Speaker: Pascal Leroy. Multi-agent reinforcement learning.
Lesson 11. 28/04/2021. Speaker: Damien Ernst. Exploration/exploitation in Reinforcement Learning: The multi-armed bandit problems. Class based on a discussion of Research paper 9 (first 25 pages). (Class not yet confirmed).
Assignment 1 – 03/02/2021. Section 1 to 4 need to be submitted for the 09/02/2021 midnight. Section 5 and 6 for the 16/02/2021 midnight. Link to the submission platform: https://submit.montefiore.ulg.ac.be/index.php/login
Results assignments 1
Assignment 2 – 17/02/2021. Section 1 to 4 need to be submitted for the 23/02/2021 midnight. Deadline for the final submission: 02/03/2021 midnight.
Assignment 3 – 17/03/2021 (final assignment). Deadline for the final submission: 27/04/2021.
3. Final project
The final project will be about designing your own intelligent agent to control a double inverted pendulum – a well-known but challenging, chaotic physical system – using any reinforcement learning algorithm which is compatible with continuous state and action spaces.
Deadline for the final report is on the 14/05/2021 midnight
Schedule for the project defense
Due to the COVD-19 crisis, the 5 evaluations that usually took place during the semester have been cancelled. But there will be an oral exam based on all the material we have seen during the class!
[Evaluations that took place during the previous years
5. Access your results
Access your results
6. Final exam
Schedule for the final exam