Fast Reinforcement Learning of Dialog Strategies

9/21/00


Click here to start


Table of Contents

Fast Reinforcement Learning of Dialog Strategies

Overview

The Dialog Problem

Previous Work

Markov Decision Processes

Reinforcement Learning

Modeling a Dialog Task

Expected Cost of Standard Dialog Strategies

Expected Cost Using Reinforcement Learning

Fast Reinforcement Learning

Using back-off information

Incorporating back-off information

Training Data Requirements

Conclusions and Future Work

Many thanks to...

 Author: Joelle Pineau

Email: jpineau@cs.cmu.edu

Home Page: www.cs.cmu.edu/~jpineau