Hybrid Supervised Reinforced Model for Dialogue Systems

TL;DR

This paper introduces a hybrid recurrent model using Deep Recurrent Q-Networks for task-oriented dialogue systems, improving performance, learning speed, and robustness by modeling interaction context in latent space.

Contribution

It proposes a novel recurrent hybrid model that effectively handles dialogue state tracking and decision making with interpretability of latent representations.

Findings

Greater performance than non-recurrent baselines

Faster learning and increased robustness

Interpretable latent interaction representations

Abstract

This paper presents a recurrent hybrid model and training procedure for task-oriented dialogue systems based on Deep Recurrent Q-Networks (DRQN). The model copes with both tasks required for Dialogue Management: State Tracking and Decision Making. It is based on modeling Human-Machine interaction into a latent representation embedding an interaction context to guide the discussion. The model achieves greater performance, learning speed and robustness than a non-recurrent baseline. Moreover, results allow interpreting and validating the policy evolution and the latent representations information-wise.