{"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/open-category-detection-with-pac-guarantees","title":"Open Category Detection with PAC Guarantees","arxiv_id":"1808.00529","date":"2018-08-01","proceeding":"ICML 2018 7","authors":["Si Liu","Risheek Garrepalli","Thomas G. Dietterich","Alan Fern","Dan Hendrycks"],"abstract":"Open category detection is the problem of detecting \"alien\" test instances\nthat belong to categories or classes that were not present in the training\ndata. In many applications, reliably detecting such aliens is central to\nensuring the safety and accuracy of test set predictions. Unfortunately, there\nare no algorithms that provide theoretical guarantees on their ability to\ndetect aliens under general assumptions. Further, while there are algorithms\nfor open category detection, there are few empirical results that directly\nreport alien detection rates. Thus, there are significant theoretical and\nempirical gaps in our understanding of open category detection. In this paper,\nwe take a step toward addressing this gap by studying a simple, but\npractically-relevant variant of open category detection. In our setting, we are\nprovided with a \"clean\" training set that contains only the target categories\nof interest and an unlabeled \"contaminated\" training set that contains a\nfraction $\\alpha$ of alien examples. Under the assumption that we know an upper\nbound on $\\alpha$, we develop an algorithm with PAC-style guarantees on the\nalien detection rate, while aiming to minimize false alarms. Empirical results\non synthetic and standard benchmark datasets demonstrate the regimes in which\nthe algorithm can be effective and provide a baseline for further advancements.","url_abs":"http://arxiv.org/abs/1808.00529v1","url_pdf":"http://arxiv.org/pdf/1808.00529v1.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":"open-category-detection-with-pac-guarantees","repo_url":"https://github.com/liusi2019/ocd","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":{"status":"ok"}}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1808.00529","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}