# Semi-supervised Acoustic Event Detection based on tri-training

## TL;DR

This paper introduces a semi-supervised acoustic event detection method based on tri-training, leveraging unlabeled data and ensemble models to improve detection accuracy, especially for rare events, and distilling this knowledge into a single efficient model.

## Contribution

The paper proposes a novel tri-training based semi-supervised approach for AED that outperforms existing methods and enables knowledge distillation into a single high-accuracy model.

## Key findings

- Tri-training improves AED accuracy over supervised and self-training methods.
- Ensemble models enhance detection performance and can be distilled into a single model.
- The distilled model maintains high accuracy comparable to ensemble teachers.

## Abstract

This paper presents our work of training acoustic event detection (AED) models using unlabeled dataset. Recent acoustic event detectors are based on large-scale neural networks, which are typically trained with huge amounts of labeled data. Labels for acoustic events are expensive to obtain, and relevant acoustic event audios can be limited, especially for rare events. In this paper we leverage an Internet-scale unlabeled dataset with potential domain shift to improve the detection of acoustic events. Based on the classic tri-training approach, our proposed method shows accuracy improvement over both the supervised training baseline, and semisupervised self-training set-up, in all pre-defined acoustic event detection tasks. As our approach relies on ensemble models, we further show the improvements can be distilled to a single model via knowledge distillation, with the resulting single student model maintaining high accuracy of teacher ensemble models.