Uncertainty Calibration for Deep Audio Classifiers

TL;DR

This paper investigates uncertainty calibration methods for deep audio classifiers, revealing that spectral-normalized Gaussian process (SNGP) performs best in providing well-calibrated predictions for environment sound and music genre classification.

Contribution

It provides an empirical comparison of calibration techniques for deep audio classifiers, highlighting the effectiveness of SNGP in this domain.

Findings

Uncalibrated models tend to be over-confident.

SNGP outperforms other calibration methods in accuracy and efficiency.

Calibration methods improve reliability of audio classification predictions.

Abstract

Although deep Neural Networks (DNNs) have achieved tremendous success in audio classification tasks, their uncertainty calibration are still under-explored. A well-calibrated model should be accurate when it is certain about its prediction and indicate high uncertainty when it is likely to be inaccurate. In this work, we investigate the uncertainty calibration for deep audio classifiers. In particular, we empirically study the performance of popular calibration methods: (i) Monte Carlo Dropout, (ii) ensemble, (iii) focal loss, and (iv) spectral-normalized Gaussian process (SNGP), on audio classification datasets. To this end, we evaluate (i-iv) for the tasks of environment sound and music genre classification. Results indicate that uncalibrated deep audio classifiers may be over-confident, and SNGP performs the best and is very efficient on the two datasets of this paper.