Multiclass threshold-based classification

TL;DR

This paper proposes a novel multiclass classification framework using multidimensional thresholds, enabling post-training score optimization and improved prediction accuracy across various networks and datasets.

Contribution

It introduces a threshold-based approach that generalizes the argmax rule, allowing for a posteriori threshold tuning and the development of score-oriented loss functions.

Findings

Threshold tuning improves classification performance.

Score-oriented losses are competitive with standard functions.

Framework generalizes binary threshold methods to multiclass.

Abstract

In this paper, we introduce a threshold-based framework for multiclass classification that generalizes the standard argmax rule. This is done by replacing the probabilistic interpretation of softmax outputs with a geometric one on the multidimensional simplex, where the classification depends on a multidimensional threshold. This change of perspective enables for any trained classification network an a posteriori optimization of the classification score by means of threshold tuning, as usually carried out in the binary setting. This allows a further refinement of the prediction capability of any network. Moreover, this multidimensional threshold-based setting makes it possible to define score-oriented losses, which are based on the interpretation of the threshold as a random variable. Our experiments show that the multidimensional threshold tuning yields consistent performance improvements across various networks and datasets, and that the proposed multiclass score-oriented losses are competitive with standard loss functions, resembling the advantages observed in the binary case.