Single Model Ensemble using Pseudo-Tags and Distinct Vectors

Ryosuke Kuwabara, Jun Suzuki, Hideki Nakayama


Abstract
Model ensemble techniques often increase task performance in neural networks; however, they require increased time, memory, and management effort. In this study, we propose a novel method that replicates the effects of a model ensemble with a single model. Our approach creates K-virtual models within a single parameter space using K-distinct pseudo-tags and K-distinct vectors. Experiments on text classification and sequence labeling tasks on several datasets demonstrate that our method emulates or outperforms a traditional model ensemble with 1/K-times fewer parameters.
Anthology ID:
2020.acl-main.271
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:
3006–3013
Language:
URL:
https://aclanthology.org/2020.acl-main.271
DOI:
10.18653/v1/2020.acl-main.271
Bibkey:
Cite (ACL):
Ryosuke Kuwabara, Jun Suzuki, and Hideki Nakayama. 2020. Single Model Ensemble using Pseudo-Tags and Distinct Vectors. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 3006–3013, Online. Association for Computational Linguistics.
Cite (Informal):
Single Model Ensemble using Pseudo-Tags and Distinct Vectors (Kuwabara et al., ACL 2020)
Copy Citation:
PDF:
https://aclanthology.org/2020.acl-main.271.pdf
Video:
 http://slideslive.com/38929422
Data
IMDb Movie ReviewsRCV1