[FDD3359] Reinforcement Learning Phd Student Level 2023

This PhD-level course is about Reinforcement Learning (RL), the third type of learning besides Supervised Learning and Unsupervised Learning. In RL, the paradigm is to try out different options and learn from feedback which options are best to reach a goal. RL has its roots both in biologically and psychologically inspired learning approaches and in control. Active participation in discussion after each lecture is expected.

This page is the course page for the 2023 version of the course. This course round will be given in a flipped classroom setting. This means that you will see recordings of the theoretical part and we will meet to discuss your questions on the matter. You can access all the course material through this course website.

The tentative schedule for our course is:

• Lecture 1 (Alex, 16.02.): Introduction to Reinforcement Learning I
• Lecture 2 (Alex, 23.02.): Introduction to Reinforcement Learning II
• Lecture 3 (Ali, 02.03.): Offline Reinforcement Learning
• Lecture 4 (Ali, 09.03.): Meta-learning in RL
• Lecture 5 (Hang, 23.03.): Data-efficiency for RL in control applications
• Lecture 6 (Alexis, 30.03.): Constrained RL with temporal logic constraints
• Lecture 7 (Alexis, 06.04.): Human-robot-interaction in RL
• Lecture 8 (Chris, 13.04.): Safe RL for Control Problems
• Lecture 9 (Chris, 20.04.): Safe RL for Control Problems
• Lecture 10 (Chris, 27.04.): Correcting RL policies using shielding
• Lecture 11 (Alex, 04.05.): Application of RL to chemistry and biology 
• Presentation Day (end of June, before Midsommar)