Paper: A Comparative Study of Reinforcement Learning Techniques on Dialogue Management

ACL ID E12-3003
Title A Comparative Study of Reinforcement Learning Techniques on Dialogue Management
Venue Annual Meeting of The European Chapter of The Association of Computational Linguistics
Session Student Session
Year 2012
Authors

Adaptive Dialogue Systems are rapidly be- coming part of our everyday lives. As they progress and adopt new technologies they become more intelligent and able to adapt better and faster to their environment. Re- search in this field is currently focused on how to achieve adaptation, and particularly on applying Reinforcement Learning (RL) techniques, so a comparative study of the related methods, such as this, is necessary. In this work we compare several standard and state of the art online RL algorithms that are used to train the dialogue manager in a dynamic environment, aiming to aid re- searchers / developers choose the appropri- ate RL algorithm for their system. This is the first work, to the best of our knowledge, to evaluate online RL algorithms on the di- alogue problem and in a ...