Tighter bounds lead to improved classifiers

TL;DR

This paper proposes updating the upper bound during classifier training to better focus on relevant examples, leading to improved classification performance and better integration with larger systems.

Contribution

It introduces a dynamic upper bound approach that enhances classifier training and system-level performance integration.

Findings

Improved classification rates with the new bound approach.

Enhanced ability to incorporate external system constraints.

Better alignment of classifier performance with overall system goals.

Abstract

The standard approach to supervised classification involves the minimization of a log-loss as an upper bound to the classification error. While this is a tight bound early on in the optimization, it overemphasizes the influence of incorrectly classified examples far from the decision boundary. Updating the upper bound during the optimization leads to improved classification rates while transforming the learning into a sequence of minimization problems. In addition, in the context where the classifier is part of a larger system, this modification makes it possible to link the performance of the classifier to that of the whole system, allowing the seamless introduction of external constraints.