This class will be given during the second semester on Wednesday afternoon in Room S.39 of Building B37. It starts at 2:00pm. The first class is on the 10 th of February.  The teaching assistant for the class is Mr. Samy Aittahar  (Email: eb.eg1576150817eilu@1576150817rahat1576150817tias1576150817 ).

Students should come to the class with their laptop.

1. Lectures

Lesson 1. 06/02/2019. Speaker: Damien Ernst.  Introduction to Reinforcement Rearning (RL). Understand how to build a RL agent for non-adversarial environment with discrete state/action spaces.

Lesson 2. 13/02/2019. Speaker: Damien Ernst.  The Q-learning algorithm (see Slides Lesson 1).

Lesson 3. 20/02/2019. Speaker: Damien Ernst. Reinforcement learning for continuous state-action spaces (see Slides Lesson 1).

Lesson 4. 27/02/2019. Speaker: Damien Ernst. Discussion  Research paper 1.

Lesson 5. 06/03/2019. Speaker: Damien Ernst.  Advanced algorithms for learning Q-functions. Discussion Research paper 2.

Lesson 6. 13/03/2019. Speaker: Damien Ernst.  Policy Gradients. Discussion Research paper 3 and Research paper 4.

Lesson 7. 20/03/2019. Speaker: Raphael Fonteneau. Advanced batch mode reinforcement learning. Discussion Research paper 5.

Lesson 8. 27/03/2019. Speaker: Nicolas Vecoven. Introducing neuromodulation in deep neural networks to learn adaptive behaviors. Discussion Research paper 6.

Lesson 9. 03/04/2019. Speaker: Damien Ernst. Exploration/exploitation in Reinforcement Learning: The multi-armed bandit problems. Class based on a discussion of  Research paper 7 (first 25 pages).

Lesson 10. 24/04/2019.  Project presentation.


2. Assignments

Assignment 1 – 06/02/2019. Section 1 to 3 need to be submitted for the 12/02/2019 midnight. Section 4 and 5 for the 19/02/2019. Deadline for the final submission: 26/02/2019 midnight. Link to the submission platform:

Results assignments 1

Assignment 2 – 27/02/2019. Preliminary results due to for the 06/03/2019 midnight. Deadline for the final submission: 27/03/2019 midnight.

Assignment 3 – /02/2019 (final assignment). Preliminary results due to for the  20/03/2019. Deadline for the final submission: 15/05/2018.  Assignment 3 is not mandatory. But if you get between 10/20-15/20 for the assignment, you get a +2 bonus on your final note, and a plus +4 bonus if you are above 15 and below 20. If you get 20/20, you get a plus 6 bonus.

3. Final project

The final project will be about designing your own intelligent agent for a video game (Starcraft II ; just kidding) using any reinforcement learning technique you want. You have to report the performances of your algorithms according to the number of episodes played by your agent. And the faster it learns, the higher your note for the project will be.

Deadline for the final report is on the 07/05/2019. Presentation of your results on the 08/07/2019.

Download your final project

Link to the source code defining the environment

Schedule for the project defense

4. Evaluations

There will be a total of 5  evaluations.

Evaluation 1 – Results

Evaluation 2 – Results

Evaluation 3 – Results

Evaluation 4 – Results

Evaluation 5

5. Final exam 

Schedule for the final exam


Share on FacebookTweet about this on TwitterShare on LinkedIn