diochnos/teaching/CS5713
CS 5713 – Computational Learning Theory
Flyer for the Course
Here you can find a flyer for the course for the fall of 2023.
Course Description
Topics of machine learning theory. Learning using membership queries, equivalence queries, version spaces, linear models, decision trees. Probably approximately correct (PAC) learning, Occam algorithms, VC-dimension, sample sizes for distribution-independent learning. Representation issues, proper learning, reductions, intractability, learning in the realizable case, agnostic learning. We explore topics under the broader umbrella of trustworthy machine learning, such as noise models, statistical queries, and adversarially robust learning against poisoning attacks and against adversarial examples. In addition, we expland upon interpretability and explainability aspects, as well as upon fairness concerns. Other topics include distribution-specific learning and evolvability, online learning and learning with expert advice in the mistake bound model, and weak and strong learning (boosting).
Times Offered
Below is a list of times when CS 5713 – Computational Learning Theory has been (or is scheduled to be) offered and taught by me.
The first three times the course was offered, was under the general CS 5970 course code that is assigned to various graduate-level seminars.
