Budgeted Policy Learning for Task-Oriented Dialogue Systems

Abstract

This paper presents a new approach that extends Deep Dyna-Q (DDQ) by incorporating a Budget-Conscious Scheduling (BCS) to best utilize a fixed, small amount of user interactions (budget) for learning task-oriented dialogue agents. BCS consists of (1) a Poisson-based global scheduler to allocate budget over different stages of training; (2) a controller to decide at each training step whether the agent is trained using real or simulated experiences; (3) a user goal sampling module to generate the experiences that are most effective for policy learning. Experiments on a movie-ticket booking task with simulated and real users show that our approach leads to significant improvements in success rate over the state-of-the-art baselines given the fixed budget.

10 pages, 7 figures, ACL 2019

References in corpus (4)

• A User Simulator for Task-Completion Dialogues
• Continuously Learning Neural Dialogue Management
• End-to-End Task-Completion Neural Dialogue Systems
• Thompson sampling with the online bootstrap

Cited by in corpus (5)

• Task-Oriented Dialogue System as Natural Language Generation
• A Survey on Dialog Management: Recent Advances and Challenges
• Multi-Agent Task-Oriented Dialog Policy Learning with Role-Aware Reward Decomposition
• Semi-Supervised Dialogue Policy Learning via Stochastic Reward Estimation
• Learning Goal-oriented Dialogue Policy with Opposite Agent Awareness