CSCI-B659 (Topics in Artificial Intelligence): 
Learning Theory and Probabilistic Graphical Models

Instructor: Roni Khardon

The course will focus on two aspects: approximation algorithms and inference in probabilistic machine learning and computational learning theory. The first topic requires B555 (or equivalent knowledge) as a pre-requisite and builds a more thorough understanding of state of the art algorithms. The second requires some basic knowledge of CS-theory and provides a framework and analysis techniques that enable understanding of when and why machine learning works. We will work toward a combination of the two, that is, a theory that explains when approximation algorithms in probabilistic machine learning succeed and how to design such algorithms.
We will start with a lecture based format to learn some foundations in these areas and then read recent papers from the literature.