{"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/online-control-of-the-false-discovery-rate","title":"Online control of the false discovery rate with decaying memory","arxiv_id":"1710.00499","date":"2017-10-02","proceeding":"NeurIPS 2017 12","authors":["Aaditya Ramdas","Fanny Yang","Martin J. Wainwright","Michael. I. Jordan"],"abstract":"In the online multiple testing problem, p-values corresponding to different\nnull hypotheses are observed one by one, and the decision of whether or not to\nreject the current hypothesis must be made immediately, after which the next\np-value is observed. Alpha-investing algorithms to control the false discovery\nrate (FDR), formulated by Foster and Stine, have been generalized and applied\nto many settings, including quality-preserving databases in science and\nmultiple A/B or multi-armed bandit tests for internet commerce. This paper\nimproves the class of generalized alpha-investing algorithms (GAI) in four\nways: (a) we show how to uniformly improve the power of the entire class of\nmonotone GAI procedures by awarding more alpha-wealth for each rejection,\ngiving a win-win resolution to a recent dilemma raised by Javanmard and\nMontanari, (b) we demonstrate how to incorporate prior weights to indicate\ndomain knowledge of which hypotheses are likely to be non-null, (c) we allow\nfor differing penalties for false discoveries to indicate that some hypotheses\nmay be more important than others, (d) we define a new quantity called the\ndecaying memory false discovery rate (mem-FDR) that may be more meaningful for\ntruly temporal applications, and which alleviates problems that we describe and\nrefer to as \"piggybacking\" and \"alpha-death\". Our GAI++ algorithms incorporate\nall four generalizations simultaneously, and reduce to more powerful variants\nof earlier algorithms when the weights and decay are all set to unity. Finally,\nwe also describe a simple method to derive new online FDR rules based on an\nestimated false discovery proportion.","url_abs":"http://arxiv.org/abs/1710.00499v1","url_pdf":"http://arxiv.org/pdf/1710.00499v1.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":"online-control-of-the-false-discovery-rate","repo_url":"https://github.com/fanny-yang/MABFDR","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok"}}],"tasks":[{"task_slug":"unity","task_name":"Unity"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":"https://app.syntology.ai/?focus=1710.00499","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}