{"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/metric-learning-for-novelty-and-anomaly","title":"Metric Learning for Novelty and Anomaly Detection","arxiv_id":"1808.05492","date":"2018-08-16","proceeding":null,"authors":["Marc Masana","Idoia Ruiz","Joan Serrat","Joost Van de Weijer","Antonio M. Lopez"],"abstract":"When neural networks process images which do not resemble the distribution\nseen during training, so called out-of-distribution images, they often make\nwrong predictions, and do so too confidently. The capability to detect\nout-of-distribution images is therefore crucial for many real-world\napplications. We divide out-of-distribution detection between novelty detection\n---images of classes which are not in the training set but are related to\nthose---, and anomaly detection ---images with classes which are unrelated to\nthe training set. By related we mean they contain the same type of objects,\nlike digits in MNIST and SVHN. Most existing work has focused on anomaly\ndetection, and has addressed this problem considering networks trained with the\ncross-entropy loss. Differently from them, we propose to use metric learning\nwhich does not have the drawback of the softmax layer (inherent to\ncross-entropy methods), which forces the network to divide its prediction power\nover the learned classes. We perform extensive experiments and evaluate both\nnovelty and anomaly detection, even in a relevant application such as traffic\nsign recognition, obtaining comparable or better results than previous works.","url_abs":"http://arxiv.org/abs/1808.05492v1","url_pdf":"http://arxiv.org/pdf/1808.05492v1.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":"metric-learning-for-novelty-and-anomaly","repo_url":"https://github.com/mmasana/OoD_Mining","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"anomaly-detection","task_name":"Anomaly Detection"},{"task_slug":"metric-learning","task_name":"Metric Learning"},{"task_slug":"novelty-detection","task_name":"Novelty Detection"},{"task_slug":"out-of-distribution-detection","task_name":"Out-of-Distribution Detection"},{"task_slug":"traffic-sign-recognition","task_name":"Traffic Sign Recognition"}],"methods":[{"method_slug":"softmax","method_name":"Softmax"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1808.05492","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}