Paper Info
Title
GALAXY: A Generative Pre-trained Model for Task-Oriented Dialog with Semi-supervised Learning and Explicit Policy Injection.
Abstract
Pre-trained models have proved to be powerful in enhancing task-oriented dialog systems. However, current pre-training methods mainly focus on enhancing dialog understanding and generation tasks while neglecting the exploitation of dialog policy. In this paper, we propose GALAXY, a novel pre-trained dialog model that explicitly learns dialog policy from limited labeled dialogs and large-scale unlabeled dialog corpora via semi-supervised learning. Specifically, we introduce a dialog act prediction task for policy optimization during pre-training and employ a consistency regularization term to refine the learned representation with the help of unlabeled dialogs. We also implement a gating mechanism to weigh suitable unlabeled dialog samples. Empirical results show that GALAXY substantially improves the performance of task-oriented dialog systems, and achieves new state-of-the-art results on benchmark datasets: In-Car, MultiWOZ2.0 and MultiWOZ2.1, improving their end-to-end combined scores by 2.5, 5.3 and 5.5 points, respectively. We also show that GALAXY has a stronger few-shot ability than existing models under various low-resource settings. For reproducibility, we release the code and data at https://github.com/siat-nlp/GALAXY.
Year
Venue
Keywords
2022
AAAI Conference on Artificial Intelligence
Speech & Natural Language Processing (SNLP)
DocType
Citations 
PageRank 
Conference
0
0.34
References 
Authors
0
12
Name
Order
Citations
PageRank
Wanwei He121.83
Yinpei Dai201.35
Yinhe Zheng313.06
Yuchuan Wu400.68
Zheng Cao500.68
Dermot Liu600.34
Peng Jiang7873.87
Min Yang87720.41
Fei Huang950656.44
Luo Si102498169.52
Jian Sun1102.70
Yongbin Li1237.49