Variational Denoising Autoencoders and Least-Squares Policy Iteration for Statistical Dialogue Managers

• Published in: IEEE signal processing letters 2020, Vol.27, p.960-964
• Main Authors: Diakoloukas, Vassilios, Lygerakis, Fotios, Lagoudakis, Michail G., Kotti, Margarita
• Format: Article
• Language: English
• Subjects: Algorithms; Approximation algorithms; Degradation; denoising; dialogue systems; Encoding; Least squares; least-squares policy iteration; Management systems; Noise; Noise reduction; Optimization; Performance degradation; sample-efficient statistical dialogue managers; Signal processing algorithms; Training; Variational autoencoders
• ISSN: 1070-9908; 1558-2361
• DOI: 10.1109/LSP.2020.2998361

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.