Frustratingly Easy Meta-Embedding – Computing Meta-Embeddings by Averaging Source Word Embeddings

Joshua Coates, Danushka Bollegala


Abstract
Creating accurate meta-embeddings from pre-trained source embeddings has received attention lately. Methods based on global and locally-linear transformation and concatenation have shown to produce accurate meta-embeddings. In this paper, we show that the arithmetic mean of two distinct word embedding sets yields a performant meta-embedding that is comparable or better than more complex meta-embedding learning methods. The result seems counter-intuitive given that vector spaces in different source embeddings are not comparable and cannot be simply averaged. We give insight into why averaging can still produce accurate meta-embedding despite the incomparability of the source vector spaces.
Anthology ID:
N18-2031
Volume:
Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 2 (Short Papers)
Month:
June
Year:
2018
Address:
New Orleans, Louisiana
Editors:
Marilyn Walker, Heng Ji, Amanda Stent
Venue:
NAACL
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
194–198
Language:
URL:
https://aclanthology.org/N18-2031
DOI:
10.18653/v1/N18-2031
Bibkey:
Cite (ACL):
Joshua Coates and Danushka Bollegala. 2018. Frustratingly Easy Meta-Embedding – Computing Meta-Embeddings by Averaging Source Word Embeddings. In Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 2 (Short Papers), pages 194–198, New Orleans, Louisiana. Association for Computational Linguistics.
Cite (Informal):
Frustratingly Easy Meta-Embedding – Computing Meta-Embeddings by Averaging Source Word Embeddings (Coates & Bollegala, NAACL 2018)
Copy Citation:
PDF:
https://aclanthology.org/N18-2031.pdf