Estimating the Spectral Moments of the Kernel Integral Operator from Finite Sample Matrices

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Chun, Chanwoo, Chung, SueYeon, Lee, Daniel D.
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866915144159723520
author Chun, Chanwoo
Chung, SueYeon
Lee, Daniel D.
author_facet Chun, Chanwoo
Chung, SueYeon
Lee, Daniel D.
contents Analyzing the structure of sampled features from an input data distribution is challenging when constrained by limited measurements in both the number of inputs and features. Traditional approaches often rely on the eigenvalue spectrum of the sample covariance matrix derived from finite measurement matrices; however, these spectra are sensitive to the size of the measurement matrix, leading to biased insights. In this paper, we introduce a novel algorithm that provides unbiased estimates of the spectral moments of the kernel integral operator in the limit of infinite inputs and features from finitely sampled measurement matrices. Our method, based on dynamic programming, is efficient and capable of estimating the moments of the operator spectrum. We demonstrate the accuracy of our estimator on radial basis function (RBF) kernels, highlighting its consistency with the theoretical spectra. Furthermore, we showcase the practical utility and robustness of our method in understanding the geometry of learned representations in neural networks.
format Preprint
id arxiv_https___arxiv_org_abs_2410_17998
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Estimating the Spectral Moments of the Kernel Integral Operator from Finite Sample Matrices
Chun, Chanwoo
Chung, SueYeon
Lee, Daniel D.
Machine Learning
Spectral Theory
Statistics Theory
Analyzing the structure of sampled features from an input data distribution is challenging when constrained by limited measurements in both the number of inputs and features. Traditional approaches often rely on the eigenvalue spectrum of the sample covariance matrix derived from finite measurement matrices; however, these spectra are sensitive to the size of the measurement matrix, leading to biased insights. In this paper, we introduce a novel algorithm that provides unbiased estimates of the spectral moments of the kernel integral operator in the limit of infinite inputs and features from finitely sampled measurement matrices. Our method, based on dynamic programming, is efficient and capable of estimating the moments of the operator spectrum. We demonstrate the accuracy of our estimator on radial basis function (RBF) kernels, highlighting its consistency with the theoretical spectra. Furthermore, we showcase the practical utility and robustness of our method in understanding the geometry of learned representations in neural networks.
title Estimating the Spectral Moments of the Kernel Integral Operator from Finite Sample Matrices
topic Machine Learning
Spectral Theory
Statistics Theory
url https://arxiv.org/abs/2410.17998