Master Computer Science, Leiden University, The Netherlands
Welcome to the webpage of the master course 'Reinforcement Learning' taught at Leiden University
Welcome to the master course "Reinforcement Learning", which will run in the spring of each year. The course consists of 14 weeks, in which you hand in 3 assignments, and make a final exam. Reinforcement Learning is an interesting topic - how can we make computers learn like animals and humans - and we try to make this course as much fun as well. The course teaches RL theory, but also has a strong practical focus, with several graded assignments about hands-on deep reinforcement learning.
Instructor: Aske Plaat
Practicals: Felix Kleuker, Koen Ponse, Alvaro Serra-Gomez, Xuening Xin, Lorenzo Madiai, Ayush Sengar, Lex Janssens, Tom van Gelooven
In-person education: The course is taught in person. Reinforcement Learning is NOT an online course!
Brightspace: We use Brightspace for all communication throughout the course: announcements, delivering slides, handing-out & handing in assignments, providing grades, content questions (forum), etc.
Attend the lectures: You should attend all lectures if you want to pass the course. The lecture slides are on Brightspace. Do not come if you are sick. Ask a friend for their lecture notes. Or re-read the chapter in the book, go over the slides, and ask questions when you have recovered.
Assignments & practicals: The assignments are the core learning experience of the course. You are encouraged to work on the assignments during the practical and ask the TAs your questions.
Read the book: In order to pass this course, you must read the book (see Literature) - lectures only is not enough. Read the chapter before you come to the lecture. Ask questions. The teachers are friendly and love answering questions.
Prior knowledge: The course is about DEEP Reinforcement Learning. We assume that you are familiar with classical supervised machine learning and with deep learning (neural networks, loss function, backpropagation, etc). In the past we have provided a special refresher summary lecture based on Appendix B of the book. The slides are still on Brightspace. We do not give this lecture anymore, due to lack of demand. Appendix B of the book may still be useful.
Note that this is not an introductory course. If your prior knowledge is lacking, better fix it before entering the course. Furthermore, although this course is listed as "mandatory" for some tracks, you can always request a program change, for example because of a mismatch in prior knowledge, from the Board of Examiners, who have been known to be quite reasonable in their decisions. We are here to help, not to make your life miserable or to force you to study topics that are not a good match. Help us help you.
A more theoretical alternative might be: Causal Inference
A more data science alternative might be: Information Retrieval
A more programming alternative might be: High Performance Computing, or Cloud Computing
A more applied alternative might be: Sports Data Science
A more machine learning alternative might be: Recommender Systems
A more Autonomous Systems alternative mught be: Robotics
A more NLP alternative might be: Text Mining
A more bio alternative might be: Bio modeling
... but please have a good look in the study guide yourself