Adversarial Learning of Task-Oriented Neural Dialog Models
This work addresses sample efficiency and reward estimation challenges in task-oriented dialog systems, though it is incremental as it builds on existing RL methods.
The authors tackled the problem of inconsistent or unavailable user feedback in reinforcement learning-based task-oriented dialog systems by proposing an adversarial learning method to estimate rewards directly from dialog samples, achieving an advanced dialog success rate in a restaurant search domain.
In this work, we propose an adversarial learning method for reward estimation in reinforcement learning (RL) based task-oriented dialog models. Most of the current RL based task-oriented dialog systems require the access to a reward signal from either user feedback or user ratings. Such user ratings, however, may not always be consistent or available in practice. Furthermore, online dialog policy learning with RL typically requires a large number of queries to users, suffering from sample efficiency problem. To address these challenges, we propose an adversarial learning method to learn dialog rewards directly from dialog samples. Such rewards are further used to optimize the dialog policy with policy gradient based RL. In the evaluation in a restaurant search domain, we show that the proposed adversarial dialog learning method achieves advanced dialog success rate comparing to strong baseline methods. We further discuss the covariate shift problem in online adversarial dialog learning and show how we can address that with partial access to user feedback.