Analysis of Feature Representations for Anomalous Sound Detection

被引:7
|
作者
Mueller, Robert [1 ]
Illium, Steffen [1 ]
Ritz, Fabian [1 ]
Schmid, Kyrill [1 ]
机构
[1] Ludwig Maximilians Univ Munchen, Mobile & Distributed Syst Grp, Munich, Germany
关键词
Anomaly Detection; Transfer Learning; Machine Health Monitoring;
D O I
10.5220/0010226800970106
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
In this work, we thoroughly evaluate the efficacy of pretrained neural networks as feature extractors for anomalous sound detection. In doing so, we leverage the knowledge that is contained in these neural networks to extract semantically rich features (representations) that serve as input to a Gaussian Mixture Model which is used as a density estimator to model normality. We compare feature extractors that were trained on data from various domains, namely: images, environmental sounds and music. Our approach is evaluated on recordings from factory machinery such as valves, pumps, sliders and fans. All of the evaluated representations outperform the autoencoder baseline with music based representations yielding the best performance in most cases. These results challenge the common assumption that closely matching the domain of the feature extractor and the downstream task results in better downstream task performance.
引用
收藏
页码:97 / 106
页数:10
相关论文
共 50 条
  • [1] ANOMALOUS SOUND DETECTION BY FEATURE-LEVEL ANOMALY SIMULATION
    Zavrtanik, Vitjan
    Marolt, Matija
    Kristan, Matej
    Skocaj, Danijel
    2024 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING, ICASSP 2024, 2024, : 1466 - 1470
  • [2] Sub-Cluster AdaCos: Learning Representations for Anomalous Sound Detection
    Wilkinghoff, Kevin
    2021 INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS (IJCNN), 2021,
  • [3] Anomalous Sound Detection Based on Sound Separation
    Shimonishi, Kanta
    Dohi, Kota
    Kawaguchi, Yohei
    INTERSPEECH 2023, 2023, : 2733 - 2737
  • [4] Optimizing Acoustic Feature Extractor for Anomalous Sound Detection Based on Neyman-Pearson Lemma
    Koizumi, Yuma
    Saito, Shoichiro
    Uematsu, Hisashi
    Harada, Noboru
    2017 25TH EUROPEAN SIGNAL PROCESSING CONFERENCE (EUSIPCO), 2017, : 698 - 702
  • [5] An unsupervised anomalous sound detection method based on similarity-driven automatic feature selection
    Zhang, Yi
    Feng, Jie
    Zhang, Qiaoling
    Hu, Junyao
    Zhang, Weiwei
    DIGITAL SIGNAL PROCESSING, 2025, 161
  • [6] HYPERBOLIC UNSUPERVISED ANOMALOUS SOUND DETECTION
    Germain, Francois G.
    Wichern, Gordon
    Le Roux, Jonathan
    2023 IEEE WORKSHOP ON APPLICATIONS OF SIGNAL PROCESSING TO AUDIO AND ACOUSTICS, WASPAA, 2023,
  • [7] Anomalous Sound Detection Based on Machine Activity Detection
    Nishida, Tomoya
    Dohi, Kota
    Endo, Takashi
    Yamamoto, Masaaki
    Kawaguchi, Yohei
    2022 30TH EUROPEAN SIGNAL PROCESSING CONFERENCE (EUSIPCO 2022), 2022, : 269 - 273
  • [8] Murmur Detection and Classification Using Signal Processing and Sound Feature Analysis
    Sivaraman, Ram
    Xiao, Yizi
    CIRCULATION, 2024, 150
  • [9] Robust Prototype Learning for Anomalous Sound Detection
    Zeng, Xiao-Min
    Song, Yan
    McLoughlin, Ian
    Liu, Lin
    Dai, Li-Rong
    INTERSPEECH 2023, 2023, : 261 - 265
  • [10] ANOMALOUS SOUND DETECTION BASED ON ATTENTION MECHANISM
    Mori, Hayato
    Tamura, Satoshi
    Hayamizu, Satoru
    29TH EUROPEAN SIGNAL PROCESSING CONFERENCE (EUSIPCO 2021), 2021, : 581 - 585