Optimization Over Zonotopes and Training Support Vector Machines

TL;DR

This paper explores the relationship between zonotopes and SVM classifiers, using the ellipsoid method to derive new theoretical insights and examining properties of soft margin C-SVMs as the parameter C increases.

Contribution

It introduces a novel connection between zonotopes and SVMs and applies the ellipsoid method to obtain new theoretical results on SVM training.

Findings

New theoretical results on training SVMs

Properties of soft margin C-SVMs as C approaches infinity

Connection between zonotopes and SVM classifiers

Abstract

We make a connection between classical polytopes called zonotopes and Support Vector Machine (SVM) classifiers. We combine this connection with the ellipsoid method to give some new theoretical results on training SVMs. We also describe some special properties of soft margin C-SVMs as parameter C goes to infinity.