{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/unsupervised-detection-of-anomalous-sound","title":"Unsupervised Detection of Anomalous Sound based on Deep Learning and the Neyman-Pearson Lemma","arxiv_id":"1810.09133","date":"2018-10-22","proceeding":null,"authors":["Yuma Koizumi","Shoichiro Saito","Hisashi Uematsum Yuta Kawachi","Noboru Harada"],"abstract":"This paper proposes a novel optimization principle and its implementation for\nunsupervised anomaly detection in sound (ADS) using an autoencoder (AE). The\ngoal of unsupervised-ADS is to detect unknown anomalous sound without training\ndata of anomalous sound. Use of an AE as a normal model is a state-of-the-art\ntechnique for unsupervised-ADS. To decrease the false positive rate (FPR), the\nAE is trained to minimize the reconstruction error of normal sounds and the\nanomaly score is calculated as the reconstruction error of the observed sound.\nUnfortunately, since this training procedure does not take into account the\nanomaly score for anomalous sounds, the true positive rate (TPR) does not\nnecessarily increase. In this study, we define an objective function based on\nthe Neyman-Pearson lemma by considering ADS as a statistical hypothesis test.\nThe proposed objective function trains the AE to maximize the TPR under an\narbitrary low FPR condition. To calculate the TPR in the objective function, we\nconsider that the set of anomalous sounds is the complementary set of normal\nsounds and simulate anomalous sounds by using a rejection sampling algorithm.\nThrough experiments using synthetic data, we found that the proposed method\nimproved the performance measures of ADS under low FPR conditions. In addition,\nwe confirmed that the proposed method could detect anomalous sounds in real\nenvironments.","url_abs":"http://arxiv.org/abs/1810.09133v1","url_pdf":"http://arxiv.org/pdf/1810.09133v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"unsupervised-detection-of-anomalous-sound","repo_url":"https://github.com/lifesailor/data-driven-predictive-maintenance","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":null}],"tasks":[{"task_slug":"anomaly-detection","task_name":"Anomaly Detection"},{"task_slug":"lemma","task_name":"LEMMA"},{"task_slug":"unsupervised-anomaly-detection","task_name":"Unsupervised Anomaly Detection"},{"task_slug":"unsupervised-anomaly-detection-in-sound","task_name":"Unsupervised Anomaly Detection In Sound"}],"methods":[{"method_slug":"ae","method_name":"AE"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}