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

2026-02-21

AI Summary

This work addresses a key limitation in existing anomaly sound detection methods, which rely on fixed neighborhood sizes for local density estimation and often violate the locality assumption by spanning cluster boundaries, thereby degrading performance. To overcome this, the authors propose an adaptive neighborhood selection mechanism that dynamically adjusts the neighborhood range by detecting distance discontinuities at cluster boundaries, enabling structure-aware density normalization. The approach integrates distance embedding, local density estimation, and a lightweight cluster-exit detection module to enhance the robustness of normalization. Extensive experiments across multiple datasets and embedding models demonstrate that the proposed method is highly robust to the choice of neighborhood size and consistently improves detection performance.

Technology Category

• Data Mining & Knowledge Management: Anomaly/Outlier Detection
• Search and Optimization: Local Search
• Machine Learning: Clustering

Application Category

• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Web Mining and Content Analysis: Normalization, clustering, classification, and summarization of Web text
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

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.

Problem

• anomalous sound detection
• local density-based normalization
• neighborhood size
• cluster boundaries
• locality assumption

Innovation

• cluster exit detection
• local density-based normalization
• anomalous sound detection
• adaptive neighborhood size
• distance discontinuity