Composition, Attention, or Both?

Ryo Yoshida, Yohei Oseki


Abstract
In this paper, we propose a novel architecture called Composition Attention Grammars (CAGs) that recursively compose subtrees into a single vector representation with a composition function, and selectively attend to previous structural information with a self-attention mechanism. We investigate whether these components—the composition function and the self-attention mechanism—can both induce human-like syntactic generalization. Specifically, we train language models (LMs) with and without these two components with the model sizes carefully controlled, and evaluate their syntactic generalization performance against six test circuits on the SyntaxGym benchmark. The results demonstrated that the composition function and the self-attention mechanism both play an important role to make LMs more human-like, and closer inspection of linguistic phenomenon implied that the composition function allowed syntactic features, but not semantic features, to percolate into subtree representations.
Anthology ID:
2022.findings-emnlp.428
Original:
2022.findings-emnlp.428v1
Version 2:
2022.findings-emnlp.428v2
Volume:
Findings of the Association for Computational Linguistics: EMNLP 2022
Month:
December
Year:
2022
Address:
Abu Dhabi, United Arab Emirates
Editors:
Yoav Goldberg, Zornitsa Kozareva, Yue Zhang
Venue:
Findings
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
5822–5834
Language:
URL:
https://aclanthology.org/2022.findings-emnlp.428
DOI:
10.18653/v1/2022.findings-emnlp.428
Bibkey:
Cite (ACL):
Ryo Yoshida and Yohei Oseki. 2022. Composition, Attention, or Both?. In Findings of the Association for Computational Linguistics: EMNLP 2022, pages 5822–5834, Abu Dhabi, United Arab Emirates. Association for Computational Linguistics.
Cite (Informal):
Composition, Attention, or Both? (Yoshida & Oseki, Findings 2022)
Copy Citation:
PDF:
https://aclanthology.org/2022.findings-emnlp.428.pdf
Video:
 https://aclanthology.org/2022.findings-emnlp.428.mp4