Tutorial on Graphical Models in Speech and Language Processing

Abstract

Graphical models are a powerful statistical methodology that can describe and help solve problems in many application domains. Recently, much research has been devoted to their application in speech and language processing. These time series have properties that create unique research problems, many of which are still open.

Graphical models offer a general and mathematically principled way to provide solutions for these problems, and to represent and compute with a wide variety of models in speech and language.

Graphical Models Tutorial Outline (pdf, 24kb)

Speaker: Jeff A. Bilmes, University of Washington, Seattle