Disentangling Hierarchical Features for Anomalous Sound Detection Under Domain Shift

2025-01-03

AI Summary

Anomalous sound detection (ASD) suffers from degraded cross-domain generalization under domain shift, primarily due to inconsistent acoustic feature distributions caused by varying machine operating conditions. Method: This paper proposes a gradient reversal–driven hierarchical feature disentanglement framework. It is the first to jointly integrate domain-relevant/domain-irrelevant feature separation with meta-information–guided (e.g., region ID, acoustic attributes) fine-grained hierarchical modeling, implemented via gradient reversal layers, hierarchical encoders, and a meta-aware attention mechanism for adversarial feature disentanglement. Results: Evaluated on the DCASE 2022 Task 2 dataset, the method achieves significant improvements in AUC and pAUC, outperforming baseline methods by over 8.2% on average under severe domain shift. It substantially enhances model robustness and generalization capability against environmental and operational condition variations.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Multiagent Systems: Adversarial Agents
• Search and Optimization: Metareasoning and Metaheuristics

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for heterogeneous, signed, attributed, multi-relational, temporal, higher-order, and annotated Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Anomalous sound detection (ASD) encounters difficulties with domain shift, where the sounds of machines in target domains differ significantly from those in source domains due to varying operating conditions. Existing methods typically employ domain classifiers to enhance detection performance, but they often overlook the influence of domain-unrelated information. This oversight can hinder the model's ability to clearly distinguish between domains, thereby weakening its capacity to differentiate normal from abnormal sounds. In this paper, we propose a Gradient Reversal-based Hierarchical feature Disentanglement (GRHD) method to address the above challenge. GRHD uses gradient reversal to separate domain-related features from domain-unrelated ones, resulting in more robust feature representations. Additionally, the method employs a hierarchical structure to guide the learning of fine-grained, domain-specific features by leveraging available metadata, such as section IDs and machine sound attributes. Experimental results on the DCASE 2022 Challenge Task 2 dataset demonstrate that the proposed method significantly improves ASD performance under domain shift.

Problem

• Anomaly Sound Detection
• Environmental Variability
• Model Generalization

Innovation

• Gradient Reversal
• Hierarchical Feature Decoupling
• Anomaly Sound Detection