Abstract
Recent works usually address Dialog policy learning DPL by training a reinforcement learning (RL) agent to determine the best dialog action. However, existing works on deep RL require a large volume of agent-user interactions to achieve acceptable performance. In this paper, we propose to make full use of the plain text knowledge from the pre-trained language model to accelerate the RL agent’s learning speed. Specifically, we design a dialog action-aware transformer encoder (DaTrans), which integrates a new fine-tuning procedure named masked last action task to encourage DaTrans to be dialog-aware and distill action-specific features. Then, DaTrans is further optimized in an RL setting with ongoing interactions and evolves through exploration in the dialog action space toward maximizing long-term accumulated rewards. The effectiveness and efficiency of the proposed model are demonstrated with both simulator evaluation and human evaluation.- Anthology ID:
- 2023.sigdial-1.12
- Volume:
- Proceedings of the 24th Annual Meeting of the Special Interest Group on Discourse and Dialogue
- Month:
- September
- Year:
- 2023
- Address:
- Prague, Czechia
- Editors:
- Svetlana Stoyanchev, Shafiq Joty, David Schlangen, Ondrej Dusek, Casey Kennington, Malihe Alikhani
- Venue:
- SIGDIAL
- SIG:
- SIGDIAL
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 142–148
- Language:
- URL:
- https://aclanthology.org/2023.sigdial-1.12
- DOI:
- 10.18653/v1/2023.sigdial-1.12
- Cite (ACL):
- Huimin Wang, Wai Chung Kwan, and Kam-Fai Wong. 2023. Dialog Action-Aware Transformer for Dialog Policy Learning. In Proceedings of the 24th Annual Meeting of the Special Interest Group on Discourse and Dialogue, pages 142–148, Prague, Czechia. Association for Computational Linguistics.
- Cite (Informal):
- Dialog Action-Aware Transformer for Dialog Policy Learning (Wang et al., SIGDIAL 2023)
- PDF:
- https://preview.aclanthology.org/ingest-acl-2023-videos/2023.sigdial-1.12.pdf