Examining the effect of whitening on static and contextualized word embeddings. Issue 3 (May 2023)

• Record Type: Journal Article
• Authors: Sasaki, Shota; Heinzerling, Benjamin; Suzuki, Jun; Inui, Kentaro
• Is Part Of: Information processing & management. Volume 60:Issue 3(2023)
• Publication Date: 2023-05
• Subjects: Static word embeddings -- Contextualized word embeddings -- Whitening -- Frequency bias; Information storage and retrieval systems -- Periodicals; Information science -- Periodicals; Systèmes d'information -- Périodiques; Sciences de l'information -- Périodiques; Information science; Information storage and retrieval systems; Periodicals; 658.4038
• DOI: 10.1016/j.ipm.2023.103272
• Languages: English
• ISSNs: 0306-4573

Abstract

Abstract: Static word embeddings (SWE) and contextualized word embeddings (CWE) are the foundation of modern natural language processing. However, these embeddings suffer from spatial bias in the form of anisotropy, which has been demonstrated to reduce their performance. A method to alleviate the anisotropy is the "whitening" transformation. Whitening is a standard method in signal processing and other areas, however, its effect on SWE and CWE is not well understood. In this study, we conduct an experiment to elucidate the effect of whitening on SWE and CWE. The results indicate that whitening predominantly removes the word frequency bias in SWE, and biases other than the word frequency bias in CWE.