Towards End-to-End Learning for Dialog State Tracking and Management using Deep Reinforcement Learning

Tiancheng Zhao; Maxine Eskenazi

Towards End-to-End Learning for Dialog State Tracking and Management using Deep Reinforcement Learning

Artificial Intelligence 2016-09-19 v2 Computation and Language Machine Learning

Authors: Tiancheng Zhao , Maxine Eskenazi

Abstract

This paper presents an end-to-end framework for task-oriented dialog systems using a variant of Deep Recurrent Q-Networks (DRQN). The model is able to interface with a relational database and jointly learn policies for both language understanding and dialog strategy. Moreover, we propose a hybrid algorithm that combines the strength of reinforcement learning and supervised learning to achieve faster learning speed. We evaluated the proposed model on a 20 Question Game conversational game simulator. Results show that the proposed method outperforms the modular-based baseline and learns a distributed representation of the latent dialog state.

Keywords

dialogue generation reinforcement learning dialogue systems

Cite

@article{arxiv.1606.02560,
  title  = {Towards End-to-End Learning for Dialog State Tracking and Management using Deep Reinforcement Learning},
  author = {Tiancheng Zhao and Maxine Eskenazi},
  journal= {arXiv preprint arXiv:1606.02560},
  year   = {2016}
}

Comments

In proceeding of SIGDIAL 2016. Added changes based-on peer review, including: 1. Added references, 2. fixed typos in text and figures, 3. added minor change to introduction

Towards End-to-End Learning for Dialog State Tracking and Management using Deep Reinforcement Learning

Abstract

Keywords

Cite

Comments

Related papers