Anomalous Sound Detection Based on Machine Activity Detection

TL;DR

This paper introduces an unsupervised method for detecting anomalous machine sounds by leveraging machine activity detection, improving detection accuracy through auxiliary task learning and ensemble methods.

Contribution

The paper proposes a novel unsupervised anomaly detection approach that uses machine activity detection as an auxiliary task to enhance detection performance.

Findings

Improved anomaly detection performance with the proposed method.

Ensemble approach further enhances detection accuracy.

Effective in distinguishing target machine sounds from background noise.

Abstract

We have developed an unsupervised anomalous sound detection method for machine condition monitoring that utilizes an auxiliary task -- detecting when the target machine is active. First, we train a model that detects machine activity by using normal data with machine activity labels and then use the activity-detection error as the anomaly score for a given sound clip if we have access to the ground-truth activity labels in the inference phase. If these labels are not available, the anomaly score is calculated through outlier detection on the embedding vectors obtained by the activity-detection model. Solving this auxiliary task enables the model to learn the difference between the target machine sounds and similar background noise, which makes it possible to identify small deviations in the target sounds. Experimental results showed that the proposed method improves the anomaly-detection performance of the conventional method complementarily by means of an ensemble.