Evaluating Classification Systems Against Soft Labels with Fuzzy Precision and Recall

TL;DR

This paper introduces fuzzy precision, recall, and F-score metrics that evaluate classification systems directly on soft labels, avoiding binarization and providing more accurate performance measures for non-binary reference data.

Contribution

The paper proposes a novel method to compute precision, recall, and F-score for soft labels, extending traditional metrics to handle non-binary reference data without quantization.

Findings

Metrics behave as expected on simple examples

Evaluation of sound event detection models demonstrates practical applicability

Metrics provide more accurate performance assessment with soft labels

Abstract

Classification systems are normally trained by minimizing the cross-entropy between system outputs and reference labels, which makes the Kullback-Leibler divergence a natural choice for measuring how closely the system can follow the data. Precision and recall provide another perspective for measuring the performance of a classification system. Non-binary references can arise from various sources, and it is often beneficial to use the soft labels for training instead of the binarized data. However, the existing definitions for precision and recall require binary reference labels, and binarizing the data can cause erroneous interpretations. We present a novel method to calculate precision, recall and F-score without quantizing the data. The proposed metrics extend the well established metrics as the definitions coincide when used with binary labels. To understand the behavior of the metrics we show simple example cases and an evaluation of different sound event detection models trained on real data with soft labels.