LM101-067: How to use Expectation Maximization to Learn Constraint Satisfaction Solutions (Rerun)

In this episode we discuss how to learn to solve constraint satisfaction inference problems. The goal of the inference process is to infer the most probable values for unobservable variables. These constraints, however, can be learned from experience. Specifically, the important machine learning method for handling unobservable components of the data using Expectation Maximization is introduced. Check it out at: www.learningmachines101.com