Human conversation is inherently complex, often spanning many different topics/domains. This makes policy learning for dialogue systems very challenging. Standard flat reinforcement learning methods do not provide an efficient framework for modelling such dialogues. In this paper, we focus on the under-explored problem of multi-domain dialogue management. First, we propose a new method for hierarchical reinforcement learning using the option framework. Next, we show that the proposed architecture learns faster and arrives at a better policy than the existing flat ones do. Moreover, we show how pretrained policies can be adapted to more complex systems with an additional set of new actions. In doing that, we show that our approach has the potential to facilitate policy optimisation for more sophisticated multi-domain dialogue systems.
@article{arxiv.1706.06210,
title = {Sub-domain Modelling for Dialogue Management with Hierarchical Reinforcement Learning},
author = {Paweł Budzianowski and Stefan Ultes and Pei-Hao Su and Nikola Mrkšić and Tsung-Hsien Wen and Iñigo Casanueva and Lina Rojas-Barahona and Milica Gašić},
journal= {arXiv preprint arXiv:1706.06210},
year = {2017}
}