Learning Representations for New Sound Classes With Continual Self-Supervised Learning

Abstract

In this paper, we work on a sound recognition system that continually incorporates new sound classes. Our main goal is to develop a framework where the model can be updated without relying on labeled data. For this purpose, we propose adopting representation learning, where an encoder is trained using unlabeled data. This learning framework enables the study and implementation of a practically relevant use case where only a small amount of the labels is available in a continual learning context. We also make the empirical observation that a similarity-based representation learning method within this framework is robust to forgetting even if no explicit mechanism against forgetting is employed. We show that this approach obtains similar performance compared to several distillation-based continual learning methods when employed on self-supervised representation learning methods.

arXiv commentAccepted to IEEE Signal Processing Letters

References in corpus (9)

• Overcoming catastrophic forgetting in neural networks
• Representation Learning with Contrastive Predictive Coding
• SpecAugment: A Simple Data Augmentation Method for Automatic Speech Recognition
• Progressive Neural Networks
• Three scenarios for continual learning
• SpeechBrain: A General-Purpose Speech Toolkit
• Continual Learning for Fake Audio Detection
• Self-Supervised Learning Aided Class-Incremental Lifelong Learning
• Self-Supervised Class Incremental Learning

Cited by in corpus (1)

• DeCoR: Defy Knowledge Forgetting by Predicting Earlier Audio Codes