A Topological Regularizer for Classifiers via Persistent Homology

TL;DR

This paper introduces a novel topological regularizer based on persistent homology to control the complexity of classifier decision boundaries, improving model simplicity without sacrificing flexibility.

Contribution

It proposes a new topological regularization method for classifiers, with an efficient gradient computation algorithm, to simplify decision boundaries by controlling topological features.

Findings

Effective in simplifying classifier boundaries

Reduces spurious topological structures

Works on synthetic and real-world datasets

Abstract

Regularization plays a crucial role in supervised learning. Most existing methods enforce a global regularization in a structure agnostic manner. In this paper, we initiate a new direction and propose to enforce the structural simplicity of the classification boundary by regularizing over its topological complexity. In particular, our measurement of topological complexity incorporates the importance of topological features (e.g., connected components, handles, and so on) in a meaningful manner, and provides a direct control over spurious topological structures. We incorporate the new measurement as a topological penalty in training classifiers. We also pro- pose an efficient algorithm to compute the gradient of such penalty. Our method pro- vides a novel way to topologically simplify the global structure of the model, without having to sacrifice too much of the flexibility of the model. We demonstrate the effectiveness of our new topological regularizer on a range of synthetic and real-world datasets.