{"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/from-zero-to-hero-cold-start-anomaly","title":"From Zero to Hero: Cold-Start Anomaly Detection","arxiv_id":"2405.20341","date":"2024-05-30","proceeding":null,"authors":["Tal Reiss","George Kour","Naama Zwerdling","Ateret Anaby-Tavor","Yedid Hoshen"],"abstract":"When first deploying an anomaly detection system, e.g., to detect out-of-scope queries in chatbots, there are no observed data, making data-driven approaches ineffective. Zero-shot anomaly detection methods offer a solution to such \"cold-start\" cases, but unfortunately they are often not accurate enough. This paper studies the realistic but underexplored cold-start setting where an anomaly detection model is initialized using zero-shot guidance, but subsequently receives a small number of contaminated observations (namely, that may include anomalies). The goal is to make efficient use of both the zero-shot guidance and the observations. We propose ColdFusion, a method that effectively adapts the zero-shot anomaly detector to contaminated observations. To support future development of this new setting, we propose an evaluation suite consisting of evaluation protocols and metrics.","url_abs":"https://arxiv.org/abs/2405.20341v1","url_pdf":"https://arxiv.org/pdf/2405.20341v1.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":"from-zero-to-hero-cold-start-anomaly","repo_url":"https://github.com/talreiss/coldfusion","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"anomaly-detection","task_name":"Anomaly Detection"},{"task_slug":"cold-start-anomaly-detection","task_name":"Cold-Start Anomaly Detection"},{"task_slug":"zero-shot-anomaly-detection","task_name":"zero-shot anomaly detection"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/cold-start-anomaly-detection-on-banking77-oos","task":"Cold-Start Anomaly Detection","dataset":"BANKING77-OOS","model":"ColdFusion","rank_in_archive_order":1,"of":3,"metrics":{"AUC 10%":"81.7"},"uses_additional_data":false},{"leaderboard":"/sota/cold-start-anomaly-detection-on-banking77-oos","task":"Cold-Start Anomaly Detection","dataset":"BANKING77-OOS","model":"ZS","rank_in_archive_order":2,"of":3,"metrics":{"AUC 10%":"78.9"},"uses_additional_data":false},{"leaderboard":"/sota/cold-start-anomaly-detection-on-banking77-oos","task":"Cold-Start Anomaly Detection","dataset":"BANKING77-OOS","model":"DN2","rank_in_archive_order":3,"of":3,"metrics":{"AUC 10%":"76.7"},"uses_additional_data":false}],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}