The Margitron: A Generalised Perceptron with Margin

TL;DR

The paper introduces the Margitron, a generalized perceptron algorithm that converges to large margin solutions and compares favorably with SVMs in experiments.

Contribution

It broadens the perceptron framework to include margin-based classifiers, providing convergence guarantees for any desired margin fraction.

Findings

Converges in finite updates to solutions with specified margin fraction

Performs comparably to decomposition SVMs on linear kernel tasks

Effective in incremental learning settings

Abstract

We identify the classical Perceptron algorithm with margin as a member of a broader family of large margin classifiers which we collectively call the Margitron. The Margitron, (despite its) sharing the same update rule with the Perceptron, is shown in an incremental setting to converge in a finite number of updates to solutions possessing any desirable fraction of the maximum margin. Experiments comparing the Margitron with decomposition SVMs on tasks involving linear kernels and 2-norm soft margin are also reported.