A Class of Random-Kernel Network Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Author: Tian, James
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914015352979456
author Tian, James
author_facet Tian, James
contents We introduce random-kernel networks, a multilayer extension of random feature models where depth is created by deterministic kernel composition and randomness enters only in the outermost layer. We prove that deeper constructions can approximate certain functions with fewer Monte Carlo samples than any shallow counterpart, establishing a depth separation theorem in sample complexity.
format Preprint
id arxiv_https___arxiv_org_abs_2509_01090
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle A Class of Random-Kernel Network Models
Tian, James
Machine Learning
Numerical Analysis
Functional Analysis
Primary 68T07. Secondary 41A25, 41A30, 46E22
We introduce random-kernel networks, a multilayer extension of random feature models where depth is created by deterministic kernel composition and randomness enters only in the outermost layer. We prove that deeper constructions can approximate certain functions with fewer Monte Carlo samples than any shallow counterpart, establishing a depth separation theorem in sample complexity.
title A Class of Random-Kernel Network Models
topic Machine Learning
Numerical Analysis
Functional Analysis
Primary 68T07. Secondary 41A25, 41A30, 46E22
url https://arxiv.org/abs/2509.01090