Spectral Path Regression: Directional Chebyshev Harmonics for Interpretable Tabular Learning

Milo Coombs · 2026

Read Spectral Path Regression: Directional Chebyshev Harmonics for Interpretable Tabular Learning by Milo Coombs (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-2604-04091.txt.

Description

Milo Coombs built spectral path regression for open tabular learning, swapping tensor-product Chebyshev bases that blow up with size. Aimed harmonic modes laid out by aim in angular space, rather than by axis index, carry the many-input structure. Cost is kept in check by choosing a few set tone vectors, or paths, and training shrinks to a single closed-form ridge solve. Tests on standard tabular benchmarks matched strong curved rivals while staying compact and fast, with learned links spelled out in closed form.

Catalog record

CREATORMilo Coombs
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE63503
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.04091

Browse the full library · About the Tower of Babel