Publications
Detailed Information
Scale-invariant representation of machine learning
Cited 2 time in
Web of Science
Cited 3 time in Scopus
- Authors
- Issue Date
- 2022-04
- Publisher
- AMER PHYSICAL SOC
- Citation
- Physical Review e, Vol.105 No.4, p. 044306
- Abstract
- The success of machine learning has resulted from its structured representation of data. Similar data have close internal representations as compressed codes for classification or emerged labels for clustering. We observe that the frequency of internal codes or labels follows power laws in both supervised and unsupervised learning models. This scale-invariant distribution implies that machine learning largely compresses frequent typical data, and simultaneously, differentiates many atypical data as outliers. In this study, we derive the process by which these power laws can naturally arise in machine learning. In terms of information theory, the scale-invariant representation corresponds to a maximally uncertain data grouping among possible representations that guarantee a given learning accuracy.
- ISSN
- 2470-0045
- Files in This Item:
- There are no files associated with this item.
- Appears in Collections:
Item View & Download Count
Items in S-Space are protected by copyright, with all rights reserved, unless otherwise indicated.