Learning from Between-class Examples for Deep Sound Recognition

TL;DR

This paper introduces Between-Class learning (BC learning) for deep sound recognition, which mixes sounds from different classes to improve model discrimination, leading to enhanced performance surpassing human levels.

Contribution

The paper proposes BC learning, a novel data mixing strategy, and develops EnvNet-v2, a new deep sound recognition network trained with BC learning.

Findings

BC learning improves recognition accuracy across various networks and datasets.

BC learning enlarges Fisher's criterion and regularizes class feature relationships.

EnvNet-v2 trained with BC learning surpasses human-level performance.

Abstract

Deep learning methods have achieved high performance in sound recognition tasks. Deciding how to feed the training data is important for further performance improvement. We propose a novel learning method for deep sound recognition: Between-Class learning (BC learning). Our strategy is to learn a discriminative feature space by recognizing the between-class sounds as between-class sounds. We generate between-class sounds by mixing two sounds belonging to different classes with a random ratio. We then input the mixed sound to the model and train the model to output the mixing ratio. The advantages of BC learning are not limited only to the increase in variation of the training data; BC learning leads to an enlargement of Fisher's criterion in the feature space and a regularization of the positional relationship among the feature distributions of the classes. The experimental results show that BC learning improves the performance on various sound recognition networks, datasets, and data augmentation schemes, in which BC learning proves to be always beneficial. Furthermore, we construct a new deep sound recognition network (EnvNet-v2) and train it with BC learning. As a result, we achieved a performance surpasses the human level.