Characterizing Continual Learning Scenarios and Strategies for Audio Analysis

TL;DR

This paper systematically evaluates continual learning strategies for audio analysis, creating a comprehensive dataset and benchmarking methods like Replay, which outperformed others in various scenarios.

Contribution

It introduces a systematic evaluation framework and a comprehensive dataset for continual learning in audio analysis, comparing multiple approaches.

Findings

Replay outperformed other methods in DCASE challenge data.

Replay achieved 70.12% accuracy in domain incremental scenario.

Replay achieved 96.98% accuracy in class incremental scenario.

Abstract

Audio analysis is useful in many application scenarios. The state-of-the-art audio analysis approaches assume the data distribution at training and deployment time will be the same. However, due to various real-life challenges, the data may encounter drift in its distribution or can encounter new classes in the late future. Thus, a one-time trained model might not perform adequately. Continual learning (CL) approaches are devised to handle such changes in data distribution. There have been a few attempts to use CL approaches for audio analysis. Yet, there is a lack of a systematic evaluation framework. In this paper, we create a comprehensive CL dataset and characterize CL approaches for audio-based monitoring tasks. We have investigated the following CL and non-CL approaches: EWC, LwF, SI, GEM, A-GEM, GDumb, Replay, Naive, Cumulative, and Joint training. The study is very beneficial for researchers and practitioners working in the area of audio analysis for developing adaptive models. We observed that Replay achieved better results than other methods in the DCASE challenge data. It achieved an accuracy of 70.12% for the domain incremental scenario and an accuracy of 96.98% for the class incremental scenario.