Negative Training for Neural Dialogue Response Generation

Tianxing He, James Glass


Abstract
Although deep learning models have brought tremendous advancements to the field of open-domain dialogue response generation, recent research results have revealed that the trained models have undesirable generation behaviors, such as malicious responses and generic (boring) responses. In this work, we propose a framework named “Negative Training” to minimize such behaviors. Given a trained model, the framework will first find generated samples that exhibit the undesirable behavior, and then use them to feed negative training signals for fine-tuning the model. Our experiments show that negative training can significantly reduce the hit rate of malicious responses, or discourage frequent responses and improve response diversity.
Anthology ID:
2020.acl-main.185
Volume:
Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics
Month:
July
Year:
2020
Address:
Online
Editors:
Dan Jurafsky, Joyce Chai, Natalie Schluter, Joel Tetreault
Venue:
ACL
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
2044–2058
Language:
URL:
https://aclanthology.org/2020.acl-main.185
DOI:
10.18653/v1/2020.acl-main.185
Bibkey:
Cite (ACL):
Tianxing He and James Glass. 2020. Negative Training for Neural Dialogue Response Generation. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 2044–2058, Online. Association for Computational Linguistics.
Cite (Informal):
Negative Training for Neural Dialogue Response Generation (He & Glass, ACL 2020)
Copy Citation:
PDF:
https://aclanthology.org/2020.acl-main.185.pdf
Video:
 http://slideslive.com/38928691
Code
 cloudygoose/negativetraining_acl2020