Towards End-to-End Learning for Dialog State Tracking and Management using Deep Reinforcement Learning
This addresses dialog state tracking and management for conversational AI, but it appears incremental as it builds on existing reinforcement learning methods with a hybrid approach.
The paper tackles the problem of task-oriented dialog systems by proposing an end-to-end framework using Deep Recurrent Q-Networks, which jointly learns language understanding and dialog strategy policies and outperforms a modular baseline on a 20 Question Game simulator.
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.