Topics: In recent years, Bayesian techniques have been applied to a number of natural language processing tasks. The aim of this course is to provide students with an understanding of the theory behind these models, and to enable them to apply these techniques in their own research. We will study Bayesian models such as Latent Dirichlet Allocation (topic models) and (Hierarchical) Dirichlet Processes and their applications to various natural language processing tasks. We will review both variational and sampling-based inference algorithms. The course will consist of a research project and a mixture of lectures and seminar-style presentations.
Objectives: To understand the foundations of Bayesian methods in NLP. To understand current literature applying these methods to NLP tasks, and to be able to use these methods in your own research.
Target audience and prerequisites: Graduate students working in NLP and related areas. Machine learning (CS446) and prior exposure to NLP (one of CS498, LING406, CS546) or approval of the instructor required.
Literature: We will read material from a number of books as well as of original papers.