Geometric Learning with Positively Decomposable Kernels

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Da Costa, Nathael, Mostajeran, Cyrus, Ortega, Juan-Pablo, Said, Salem
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909273074696192
author Da Costa, Nathael
Mostajeran, Cyrus
Ortega, Juan-Pablo
Said, Salem
author_facet Da Costa, Nathael
Mostajeran, Cyrus
Ortega, Juan-Pablo
Said, Salem
contents Kernel methods are powerful tools in machine learning. Classical kernel methods are based on positive-definite kernels, which map data spaces into reproducing kernel Hilbert spaces (RKHS). For non-Euclidean data spaces, positive-definite kernels are difficult to come by. In this case, we propose the use of reproducing kernel Krein space (RKKS) based methods, which require only kernels that admit a positive decomposition. We show that one does not need to access this decomposition in order to learn in RKKS. We then investigate the conditions under which a kernel is positively decomposable. We show that invariant kernels admit a positive decomposition on homogeneous spaces under tractable regularity assumptions. This makes them much easier to construct than positive-definite kernels, providing a route for learning with kernels for non-Euclidean data. By the same token, this provides theoretical foundations for RKKS-based methods in general.
format Preprint
id arxiv_https___arxiv_org_abs_2310_13821
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Geometric Learning with Positively Decomposable Kernels
Da Costa, Nathael
Mostajeran, Cyrus
Ortega, Juan-Pablo
Said, Salem
Machine Learning
Differential Geometry
Kernel methods are powerful tools in machine learning. Classical kernel methods are based on positive-definite kernels, which map data spaces into reproducing kernel Hilbert spaces (RKHS). For non-Euclidean data spaces, positive-definite kernels are difficult to come by. In this case, we propose the use of reproducing kernel Krein space (RKKS) based methods, which require only kernels that admit a positive decomposition. We show that one does not need to access this decomposition in order to learn in RKKS. We then investigate the conditions under which a kernel is positively decomposable. We show that invariant kernels admit a positive decomposition on homogeneous spaces under tractable regularity assumptions. This makes them much easier to construct than positive-definite kernels, providing a route for learning with kernels for non-Euclidean data. By the same token, this provides theoretical foundations for RKKS-based methods in general.
title Geometric Learning with Positively Decomposable Kernels
topic Machine Learning
Differential Geometry
url https://arxiv.org/abs/2310.13821