Semi-Supervised Audio Classification with Partially Labeled Data

TL;DR

This paper introduces two semi-supervised learning methods for audio classification that effectively utilize partially labeled datasets, addressing the challenge of missing labels and improving model performance without requiring full annotations.

Contribution

It proposes a label enhancement approach with a teacher-student process and applies the mean teacher algorithm to handle partially labeled audio data, advancing semi-supervised techniques.

Findings

Proper handling of missing labels significantly improves classification accuracy.

Using semi-supervised methods with partially labeled data outperforms traditional supervised approaches.

Different strategies leveraging limited labels offer varying benefits depending on data conditions.

Abstract

Audio classification has seen great progress with the increasing availability of large-scale datasets. These large datasets, however, are often only partially labeled as collecting full annotations is a tedious and expensive process. This paper presents two semi-supervised methods capable of learning with missing labels and evaluates them on two publicly available, partially labeled datasets. The first method relies on label enhancement by a two-stage teacher-student learning process, while the second method utilizes the mean teacher semi-supervised learning algorithm. Our results demonstrate the impact of improperly handling missing labels and compare the benefits of using different strategies leveraging data with few labels. Methods capable of learning with partially labeled data have the potential to improve models for audio classification by utilizing even larger amounts of data without the need for complete annotations.