Feature Understanding and Sparsity Enhancement via 2-Layered kernel machines (2L-FUSE)

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Camattari, Fabiana, Guastavino, Sabrina, Marchetti, Francesco, Perracchione, Emma
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912578867822592
author Camattari, Fabiana
Guastavino, Sabrina
Marchetti, Francesco
Perracchione, Emma
author_facet Camattari, Fabiana
Guastavino, Sabrina
Marchetti, Francesco
Perracchione, Emma
contents We propose a novel sparsity enhancement strategy for regression tasks, based on learning a data-adaptive kernel metric, i.e., a shape matrix, through 2-Layered kernel machines. The resulting shape matrix, which defines a Mahalanobis-type deformation of the input space, is then factorized via an eigen-decomposition, allowing us to identify the most informative directions in the space of features. This data-driven approach provides a flexible, interpretable and accurate feature reduction scheme. Numerical experiments on synthetic and applications to real datasets of geomagnetic storms demonstrate that our approach achieves minimal yet highly informative feature sets without losing predictive performance.
format Preprint
id arxiv_https___arxiv_org_abs_2509_07806
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Feature Understanding and Sparsity Enhancement via 2-Layered kernel machines (2L-FUSE)
Camattari, Fabiana
Guastavino, Sabrina
Marchetti, Francesco
Perracchione, Emma
Numerical Analysis
Solar and Stellar Astrophysics
Machine Learning
65D15, 41A05, 68Q32
We propose a novel sparsity enhancement strategy for regression tasks, based on learning a data-adaptive kernel metric, i.e., a shape matrix, through 2-Layered kernel machines. The resulting shape matrix, which defines a Mahalanobis-type deformation of the input space, is then factorized via an eigen-decomposition, allowing us to identify the most informative directions in the space of features. This data-driven approach provides a flexible, interpretable and accurate feature reduction scheme. Numerical experiments on synthetic and applications to real datasets of geomagnetic storms demonstrate that our approach achieves minimal yet highly informative feature sets without losing predictive performance.
title Feature Understanding and Sparsity Enhancement via 2-Layered kernel machines (2L-FUSE)
topic Numerical Analysis
Solar and Stellar Astrophysics
Machine Learning
65D15, 41A05, 68Q32
url https://arxiv.org/abs/2509.07806