Unsupervised linear discrimination using skewness Journal: Journal of Multivariate Analysis
Una Radojičić, Klaus Nordhausen, Joni Virta · 2025
Read Unsupervised linear discrimination using skewness Journal: Journal of Multivariate Analysis by Una Radojičić, Klaus Nordhausen, Joni Virta (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-02412.txt.
Description
Una Radojičić, Klaus Nordhausen, and Joni Virta sort data into two groups without any labels, by exploiting lopsidedness, in this 2025 paper. When two groups form a Gaussian mixture, the direction that best separates them can be estimated blind, a classic but scattered result. Skewness, the asymmetry of the data cloud, points the way.
The authors gather four such estimators, two of them new, and derive the long run distribution of each. A neat theorem simplifies comparison: every estimator that respects affine changes of coordinates shares proportional accuracy, differing only by a constant. Simulations confirm the theory and show how the estimators behave in modest samples. The paper gives practitioners a clean menu for a task that arises wherever group labels are missing.
Catalog record
| CREATOR | Una Radojičić, Klaus Nordhausen, Joni Virta |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 123 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2508-02412