On the Consistency of the Bootstrap Approach for Support Vector Machines and Related Kernel Based Methods

TL;DR

This paper proves that bootstrap methods can reliably approximate the distribution of support vector machines and related kernel methods, providing a theoretical foundation for their use in statistical inference.

Contribution

It establishes the consistency of bootstrap approximations for SVMs with general convex loss functions and kernels, a novel theoretical result.

Findings

Bootstrap approximations are consistent for SVMs.

The result applies to general convex loss functions.

Supports the use of bootstrap for statistical inference in kernel methods.

Abstract

It is shown that bootstrap approximations of support vector machines (SVMs) based on a general convex and smooth loss function and on a general kernel are consistent. This result is useful to approximate the unknown finite sample distribution of SVMs by the bootstrap approach.