CSC321: Neural Networks
 
Lecture 16: Hidden Markov Models

What does “Markov” mean

Two ways to represent the conditional probability table of a first-order Markov process

The probability of generating a string

Learning the conditional probability table

How to have long-term dependencies and still be first order Markov

A hidden Markov model

Generating from an HMM

The probability of generating a hidden sequence

The joint probability of generating a hidden sequence and a visible sequence

The probability of generating a visible sequence from an HMM

The HMM dynamic programming trick

Learning the parameters of an HMM