Guided learning for weakly-labeled semi-supervised sound event detection

TL;DR

This paper introduces Guided Learning, a semi-supervised approach for sound event detection that uses a teacher-student model setup to improve boundary detection by leveraging weak labels and audio tagging performance.

Contribution

The paper presents a novel teacher-student framework that separates audio tagging and boundary detection, improving semi-supervised sound event detection without complex trade-offs.

Findings

Achieves competitive performance on DCASE2018 Task4 dataset

Effectively leverages unlabeled data for boundary detection

Demonstrates the benefit of separating sub-tasks in semi-supervised learning

Abstract

We propose a simple but efficient method termed Guided Learning for weakly-labeled semi-supervised sound event detection (SED). There are two sub-targets implied in weakly-labeled SED: audio tagging and boundary detection. Instead of designing a single model by considering a trade-off between the two sub-targets, we design a teacher model aiming at audio tagging to guide a student model aiming at boundary detection to learn using the unlabeled data. The guidance is guaranteed by the audio tagging performance gap of the two models. In the meantime, the student model liberated from the trade-off is able to provide more excellent boundary detection results. We propose a principle to design such two models based on the relation between the temporal compression scale and the two sub-targets. We also propose an end-to-end semi-supervised learning process for these two models to enable their abilities to rise alternately. Experiments on the DCASE2018 Task4 dataset show that our approach achieves competitive performance.