On the Spectral Structure and Objective Equivalence of Orthogonal Multilabel Fisher Discriminants
Brian Keith-Norambuena [email protected], Juan Bekios-Calfa [email protected] · 2026
Read On the Spectral Structure and Objective Equivalence of Orthogonal Multilabel Fisher Discriminants by Brian Keith-Norambuena [email protected], Juan Bekios-Calfa [email protected] (2026) 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-2605-03283.txt.
Introduction
Brian Keith-Norambuena and Juan Bekios-Calfa give a unified theory of multilabel Fisher discriminant analysis. Orthogonality constraints apply throughout. When each item carries several labels, the classic bound on discriminant directions no longer holds. Competing objective forms also need careful comparison.
They show the between-class scatter rank can exceed the single-label limit. All four Fisher objectives agree under one normalization, but diverge under the Stiefel constraint. They bound how label distances survive projection. Finite-sample error bounds come with a matching lower bound, near optimal overall. Synthetic tests check the algebra and the quantities that drive the bounds.
Catalog record
| CREATOR | Brian Keith-Norambuena [email protected], Juan Bekios-Calfa [email protected] |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 184 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.03283