Hidden Markov Models

Overview

This book build upon the use of Hidden Markov Models as motion models, which, as we have seen in chapter 3, are probably the most popular technique for pattern based motion prediction. This chapter provides the reader with a broad introduction to this probabilistic framework. Sections 4.2 through 4.4 present what may be considered as the ”classic” theory on HMMs. Readers already familiar with HMMs may safely skip these sections and proceed directly to section 4.5 and the rest of this chapter, where we discuss structure learning, a difficult problem for which no standard algorithm exists yet, despite the existence of several approaches in the literature.