Mind the Gap: Detecting Cluster Exits for Robust Local Density-Based Score Normalization in Anomalous Sound Detection

TL;DR

This paper introduces a method for adaptive neighborhood size selection in local density-based anomaly detection, improving robustness and performance across various models and datasets by detecting cluster exits.

Contribution

It proposes cluster exit detection, a lightweight technique to adapt neighborhood sizes based on local data structure, enhancing anomaly detection robustness.

Findings

Improved robustness to neighborhood size selection.

Consistent performance gains across multiple datasets.

Effective detection of cluster boundaries for better normalization.

Abstract

Local density-based score normalization is an effective component of distance-based embedding methods for anomalous sound detection, particularly when data densities vary across conditions or domains. In practice, however, performance depends strongly on neighborhood size. Increasing it can degrade detection accuracy when neighborhood expansion crosses cluster boundaries, violating the locality assumption of local density estimation. This observation motivates adapting the neighborhood size based on locality preservation rather than fixing it in advance. We realize this by proposing cluster exit detection, a lightweight mechanism that identifies distance discontinuities and selects neighborhood sizes accordingly. Experiments across multiple embedding models and datasets show improved robustness to neighborhood-size selection and consistent performance gains.