Feature Elimination in Kernel Machines in moderately high dimensions

TL;DR

This paper introduces a recursive feature elimination method for kernel machines, providing theoretical guarantees and demonstrating its effectiveness through case studies and simulations in moderately high-dimensional settings.

Contribution

It develops a new recursive feature elimination approach for kernel machines with proven consistency and practical validation.

Findings

Method is uniformly consistent in identifying the correct feature space.

Case studies show assumptions are met in practical scenarios.

Simulations demonstrate the approach's strong performance.

Abstract

We develop an approach for feature elimination in statistical learning with kernel machines, based on recursive elimination of features.We present theoretical properties of this method and show that it is uniformly consistent in finding the correct feature space under certain generalized assumptions.We present four case studies to show that the assumptions are met in most practical situations and present simulation results to demonstrate performance of the proposed approach.