{"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/active-anomaly-detection-via-ensembles","title":"Active Anomaly Detection via Ensembles","arxiv_id":"1809.06477","date":"2018-09-17","proceeding":null,"authors":["Shubhomoy Das","Md. Rakibul Islam","Nitthilan Kannappan Jayakodi","Janardhan Rao Doppa"],"abstract":"In critical applications of anomaly detection including computer security and\nfraud prevention, the anomaly detector must be configurable by the analyst to\nminimize the effort on false positives. One important way to configure the\nanomaly detector is by providing true labels for a few instances. We study the\nproblem of label-efficient active learning to automatically tune anomaly\ndetection ensembles and make four main contributions. First, we present an\nimportant insight into how anomaly detector ensembles are naturally suited for\nactive learning. This insight allows us to relate the greedy querying strategy\nto uncertainty sampling, with implications for label-efficiency. Second, we\npresent a novel formalism called compact description to describe the discovered\nanomalies and show that it can also be employed to improve the diversity of the\ninstances presented to the analyst without loss in the anomaly discovery rate.\nThird, we present a novel data drift detection algorithm that not only detects\nthe drift robustly, but also allows us to take corrective actions to adapt the\ndetector in a principled manner. Fourth, we present extensive experiments to\nevaluate our insights and algorithms in both batch and streaming settings. Our\nresults show that in addition to discovering significantly more anomalies than\nstate-of-the-art unsupervised baselines, our active learning algorithms under\nthe streaming-data setup are competitive with the batch setup.","url_abs":"http://arxiv.org/abs/1809.06477v1","url_pdf":"http://arxiv.org/pdf/1809.06477v1.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":"active-anomaly-detection-via-ensembles","repo_url":"https://github.com/shubhomoydas/ad_examples","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":null},{"paper_slug":"active-anomaly-detection-via-ensembles","repo_url":"https://github.com/freedombenLiu/ad_examples","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"active-learning","task_name":"Active Learning"},{"task_slug":"anomaly-detection","task_name":"Anomaly Detection"},{"task_slug":"computer-security","task_name":"Computer Security"},{"task_slug":"drift-detection","task_name":"Drift Detection"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}