IsoScore: Measuring the Uniformity of Embedding Space Utilization

William Rudman, Nate Gillman, Taylor Rayne, Carsten Eickhoff


Abstract
The recent success of distributed word representations has led to an increased interest in analyzing the properties of their spatial distribution. Several studies have suggested that contextualized word embedding models do not isotropically project tokens into vector space. However, current methods designed to measure isotropy, such as average random cosine similarity and the partition score, have not been thoroughly analyzed and are not appropriate for measuring isotropy. We propose IsoScore: a novel tool that quantifies the degree to which a point cloud uniformly utilizes the ambient vector space. Using rigorously designed tests, we demonstrate that IsoScore is the only tool available in the literature that accurately measures how uniformly distributed variance is across dimensions in vector space. Additionally, we use IsoScore to challenge a number of recent conclusions in the NLP literature that have been derived using brittle metrics of isotropy. We caution future studies from using existing tools to measure isotropy in contextualized embedding space as resulting conclusions will be misleading or altogether inaccurate.
Anthology ID:
2022.findings-acl.262
Volume:
Findings of the Association for Computational Linguistics: ACL 2022
Month:
May
Year:
2022
Address:
Dublin, Ireland
Editors:
Smaranda Muresan, Preslav Nakov, Aline Villavicencio
Venue:
Findings
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
3325–3339
Language:
URL:
https://aclanthology.org/2022.findings-acl.262
DOI:
10.18653/v1/2022.findings-acl.262
Bibkey:
Cite (ACL):
William Rudman, Nate Gillman, Taylor Rayne, and Carsten Eickhoff. 2022. IsoScore: Measuring the Uniformity of Embedding Space Utilization. In Findings of the Association for Computational Linguistics: ACL 2022, pages 3325–3339, Dublin, Ireland. Association for Computational Linguistics.
Cite (Informal):
IsoScore: Measuring the Uniformity of Embedding Space Utilization (Rudman et al., Findings 2022)
Copy Citation:
PDF:
https://aclanthology.org/2022.findings-acl.262.pdf
Video:
 https://aclanthology.org/2022.findings-acl.262.mp4
Code
 bcbi-edu/p_eickhoff_isoscore
Data
WikiText-2