Local Density-Based Anomaly Score Normalization for Domain Generalization

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Wilkinghoff, Kevin, Yang, Haici, Ebbers, Janek, Germain, François G., Wichern, Gordon, Roux, Jonathan Le
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912673378074624
author Wilkinghoff, Kevin
Yang, Haici
Ebbers, Janek
Germain, François G.
Wichern, Gordon
Roux, Jonathan Le
author_facet Wilkinghoff, Kevin
Yang, Haici
Ebbers, Janek
Germain, François G.
Wichern, Gordon
Roux, Jonathan Le
contents State-of-the-art anomalous sound detection (ASD) systems in domain-shifted conditions rely on projecting audio signals into an embedding space and using distance-based outlier detection to compute anomaly scores. One of the major difficulties to overcome is the so-called domain mismatch between the anomaly score distributions of a source domain and a target domain that differ acoustically and in terms of the amount of training data provided. A decision threshold that is optimal for one domain may be highly sub-optimal for the other domain and vice versa. This significantly degrades the performance when only using a single decision threshold, as is required when generalizing to multiple data domains that are possibly unseen during training while still using the same trained ASD system as in the source domain. To reduce this mismatch between the domains, we propose a simple local-density-based anomaly score normalization scheme. In experiments conducted on several ASD datasets, we show that the proposed normalization scheme consistently improves performance for various types of embedding-based ASD systems and yields better results than existing anomaly score normalization approaches.
format Preprint
id arxiv_https___arxiv_org_abs_2509_10951
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Local Density-Based Anomaly Score Normalization for Domain Generalization
Wilkinghoff, Kevin
Yang, Haici
Ebbers, Janek
Germain, François G.
Wichern, Gordon
Roux, Jonathan Le
Audio and Speech Processing
Sound
State-of-the-art anomalous sound detection (ASD) systems in domain-shifted conditions rely on projecting audio signals into an embedding space and using distance-based outlier detection to compute anomaly scores. One of the major difficulties to overcome is the so-called domain mismatch between the anomaly score distributions of a source domain and a target domain that differ acoustically and in terms of the amount of training data provided. A decision threshold that is optimal for one domain may be highly sub-optimal for the other domain and vice versa. This significantly degrades the performance when only using a single decision threshold, as is required when generalizing to multiple data domains that are possibly unseen during training while still using the same trained ASD system as in the source domain. To reduce this mismatch between the domains, we propose a simple local-density-based anomaly score normalization scheme. In experiments conducted on several ASD datasets, we show that the proposed normalization scheme consistently improves performance for various types of embedding-based ASD systems and yields better results than existing anomaly score normalization approaches.
title Local Density-Based Anomaly Score Normalization for Domain Generalization
topic Audio and Speech Processing
Sound
url https://arxiv.org/abs/2509.10951