Institutional Repository [SANDBOX]
Technical University of Crete
EN  |  EL

Search

Browse

My Space

Variational denoising autoencoders and least-squares policy iteration for statistical dialogue managers

Diakoloukas Vasileios, Lygerakis Fotios, Lagoudakis Michail, Kotti Margarita

Full record


URI: http://purl.tuc.gr/dl/dias/A1FB74B4-8B64-46FA-B0AE-941918F57F06
Year 2020
Type of Item Peer-Reviewed Journal Publication
License
Details
Bibliographic Citation V. Diakoloukas, F. Lygerakis, M. G. Lagoudakis, and M. Kotti, “Variational denoising autoencoders and least-squares policy iteration for statistical dialogue managers,” IEEE Signal Process. Lett., vol. 27, pp. 960–964, 2020. doi: 10.1109/LSP.2020.2998361 https://doi.org/10.1109/LSP.2020.2998361
Appears in Collections

Summary

The use of Reinforcement Learning (RL) approaches for dialogue policy optimization has been the new trend for dialogue management systems. Several methods have been proposed, which are trained on dialogue data to provide optimal system response. However, most of these approaches exhibit performance degradation in the presence of noise, poor scalability to other domains, as well as performance instabilities. To overcome these problems, we propose a novel approach based on the incremental, sample-efficient Least-Squares Policy Iteration (LSPI) algorithm, which is trained on compact, fixed-size dialogue state encodings, obtained from deep Variational Denoising Autoencoders (VDAE). The proposed scheme exhibits stable and noise-robust performance, which significantly outperforms the current state-of-the-art, even in mismatched noise environments.

Services

Statistics