Towards Robust Few-shot Class Incremental Learning in Audio Classification using Contrastive Representation

TL;DR

This paper introduces a contrastive learning approach to improve few-shot class-incremental audio classification, achieving state-of-the-art results by enhancing representation discriminability and generalization.

Contribution

It proposes using supervised contrastive learning during base training to better handle incremental class addition in audio classification tasks.

Findings

Achieves state-of-the-art performance on NSynth and LibriSpeech datasets.

Enhances discriminative power of audio representations.

Improves generalization in incremental learning scenarios.

Abstract

In machine learning applications, gradual data ingress is common, especially in audio processing where incremental learning is vital for real-time analytics. Few-shot class-incremental learning addresses challenges arising from limited incoming data. Existing methods often integrate additional trainable components or rely on a fixed embedding extractor post-training on base sessions to mitigate concerns related to catastrophic forgetting and the dangers of model overfitting. However, using cross-entropy loss alone during base session training is suboptimal for audio data. To address this, we propose incorporating supervised contrastive learning to refine the representation space, enhancing discriminative power and leading to better generalization since it facilitates seamless integration of incremental classes, upon arrival. Experimental results on NSynth and LibriSpeech datasets with 100 classes, as well as ESC dataset with 50 and 10 classes, demonstrate state-of-the-art performance.