Learning To Search

Learning to search is a method for solving complex joint prediction problems based on learning to search through a problem-defined search space.

A tutorial was presented ICML 2015 with slides and source by Hal Daumé and John Langford

A different related tutorial was presented at NAACL 2015 slides 1 slides 2 slides 3 with source by Hal Daumé, John Langford, Kai-Wei Chang, He He, and Sudha Rao.

Feel free to use and adapt the source as needed for any classes.