{"id":1731,"date":"2021-03-16T08:22:29","date_gmt":"2021-03-16T07:22:29","guid":{"rendered":"https:\/\/www.uni.lu\/fstm-fr\/events\/false-discovery-rate-control-with-unknown-null-distribution\/"},"modified":"2021-03-16T08:22:29","modified_gmt":"2021-03-16T07:22:29","slug":"false-discovery-rate-control-with-unknown-null-distribution","status":"publish","type":"events","link":"https:\/\/www.uni.lu\/fstm-fr\/events\/false-discovery-rate-control-with-unknown-null-distribution\/","title":{"rendered":"False Discovery rate control with unknown null distribution"},"content":{"rendered":"<section class=\"wp-block-unilux-blocks-free-section section\"><div class=\"container xl:max-w-screen-xl\"><p>Classical multiple testing theory prescribes the null distribution, which is often a too stringent assumption for nowadays large scale experiments. This paper presents theoretical foundations to understand the limitations caused by ignoring the null distribution, and how it can be properly learned from the (same) data-set, when possible. While an oracle procedure in that case is the Benjamini Hochberg procedure applied with the true (unknown) null distribution, we pursue the aim of building a procedure that asymptotically mimics the performance of the oracle (AMO in short). For a Gaussian null, our main result states that an AMO procedure exists if and only if the sparsity parameter k (number of false nulls) is of order less than n\/log(n), where n is the total number of tests.<\/p><p>This is a joint work with Nicolas Verzelen, <a href=\"https:\/\/arxiv.org\/abs\/1912.03109\" target=\"_self\" title=\"\" rel=\"noopener\">https:\/\/arxiv.org\/abs\/1912.03109<\/a>.<\/p><\/div><\/section>","protected":false},"excerpt":{"rendered":"<p>Classical multiple testing theory prescribes the null distribution, which is often a too stringent assumption for nowadays large scale experiments. This paper presents theoretical foundations to understand the limitations caused by ignoring the null distribution, and how it can be properly learned from the (same) data-set, when possible. While an oracle procedure in that case is the Benjamini Hochberg procedure applied with the true (unknown) null distribution, we pursue the aim of building a procedure that asymptotically mimics the performance of the oracle (AMO in short). For a Gaussian null, our main result states that an AMO procedure exists if and only if the sparsity parameter k (number of false nulls) is of order less than n\/log(n), where n is the total number of tests.<\/p>\n","protected":false},"author":0,"featured_media":1732,"parent":0,"menu_order":0,"comment_status":"open","ping_status":"closed","template":"","format":"standard","meta":{"featured_image_focal_point":[],"show_featured_caption":false,"ulux_newsletter_groups":"","uluxPostTitle":"","uluxPrePostTitle":"","_trash_the_other_posts":false,"_price":"","_stock":"","_tribe_ticket_header":"","_tribe_default_ticket_provider":"","_tribe_ticket_capacity":"0","_ticket_start_date":"","_ticket_end_date":"","_tribe_ticket_show_description":"","_tribe_ticket_show_not_going":false,"_tribe_ticket_use_global_stock":"","_tribe_ticket_global_stock_level":"","_global_stock_mode":"","_global_stock_cap":"","_tribe_rsvp_for_event":"","_tribe_ticket_going_count":"","_tribe_ticket_not_going_count":"","_tribe_tickets_list":"[]","_tribe_ticket_has_attendee_info_fields":false,"event_start_date":"2021-03-18 13:00:00","event_end_date":"2021-03-18 14:00:00","event_speaker_name":"Etienne Roquain from LPSM, Sorbonne University ","event_speaker_link":"","event_is_online":false,"event_location":"","event_street":"","event_location_link":"","event_zip_code":"","event_city":"","event_country":"LU"},"events-topic":[309],"events-type":[],"organisation":[60],"authorship":[],"acf":[],"yoast_head":"<!-- This site is optimized with the Yoast SEO Premium plugin v22.3 (Yoast SEO v22.3) - https:\/\/yoast.com\/wordpress\/plugins\/seo\/ -->\n<title>False Discovery rate control with unknown null distribution - FSTM I Uni.lu<\/title>\n<meta name=\"description\" content=\"Classical multiple testing theory prescribes the null distribution, which is often a too stringent assumption for nowadays large scale experiments. This paper presents theoretical foundations to understand the limitations caused by ignoring the null distribution, and how it can be properly learned from the (same) data-set, when possible. While an oracle procedure in that case is the Benjamini Hochberg procedure applied with the true (unknown) null distribution, we pursue the aim of building a procedure that asymptotically mimics the performance of the oracle (AMO in short). For a Gaussian null, our main result states that an AMO procedure exists if and only if the sparsity parameter k (number of false nulls) is of order less than n\/log(n), where n is the total number of tests.\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/www.uni.lu\/fstm-fr\/events\/false-discovery-rate-control-with-unknown-null-distribution\/\" \/>\n<meta property=\"og:locale\" content=\"fr_FR\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"False Discovery rate control with unknown null distribution\" \/>\n<meta property=\"og:description\" content=\"Classical multiple testing theory prescribes the null distribution, which is often a too stringent assumption for nowadays large scale experiments. This paper presents theoretical foundations to understand the limitations caused by ignoring the null distribution, and how it can be properly learned from the (same) data-set, when possible. While an oracle procedure in that case is the Benjamini Hochberg procedure applied with the true (unknown) null distribution, we pursue the aim of building a procedure that asymptotically mimics the performance of the oracle (AMO in short). For a Gaussian null, our main result states that an AMO procedure exists if and only if the sparsity parameter k (number of false nulls) is of order less than n\/log(n), where n is the total number of tests.\" \/>\n<meta property=\"og:url\" content=\"https:\/\/www.uni.lu\/fstm-fr\/events\/false-discovery-rate-control-with-unknown-null-distribution\/\" \/>\n<meta property=\"og:site_name\" content=\"FSTM FR\" \/>\n<meta property=\"article:publisher\" content=\"https:\/\/www.facebook.com\/fstm.uni.lu\/\" \/>\n<meta property=\"og:image\" content=\"https:\/\/www.uni.lu\/wp-content\/uploads\/sites\/20\/2026\/03\/03111744\/FSTM_SM-Profile_1600x1600px-scaled.jpg\" \/>\n\t<meta property=\"og:image:width\" content=\"2560\" \/>\n\t<meta property=\"og:image:height\" content=\"2560\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/jpeg\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"Dur\u00e9e de lecture estim\u00e9e\" \/>\n\t<meta name=\"twitter:data1\" content=\"1 minute\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\/\/www.uni.lu\/fstm-fr\/events\/false-discovery-rate-control-with-unknown-null-distribution\/\",\"url\":\"https:\/\/www.uni.lu\/fstm-fr\/events\/false-discovery-rate-control-with-unknown-null-distribution\/\",\"name\":\"False Discovery rate control with unknown null distribution - FSTM I Uni.lu\",\"isPartOf\":{\"@id\":\"https:\/\/www.uni.lu\/fstm-fr\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\/\/www.uni.lu\/fstm-fr\/events\/false-discovery-rate-control-with-unknown-null-distribution\/#primaryimage\"},\"image\":{\"@id\":\"https:\/\/www.uni.lu\/fstm-fr\/events\/false-discovery-rate-control-with-unknown-null-distribution\/#primaryimage\"},\"thumbnailUrl\":\"https:\/\/www.uni.lu\/wp-content\/uploads\/sites\/20\/2021\/03\/default-4.jpg\",\"datePublished\":\"2021-03-16T07:22:29+00:00\",\"dateModified\":\"2021-03-16T07:22:29+00:00\",\"description\":\"Classical multiple testing theory prescribes the null distribution, which is often a too stringent assumption for nowadays large scale experiments. This paper presents theoretical foundations to understand the limitations caused by ignoring the null distribution, and how it can be properly learned from the (same) data-set, when possible. While an oracle procedure in that case is the Benjamini Hochberg procedure applied with the true (unknown) null distribution, we pursue the aim of building a procedure that asymptotically mimics the performance of the oracle (AMO in short). For a Gaussian null, our main result states that an AMO procedure exists if and only if the sparsity parameter k (number of false nulls) is of order less than n\/log(n), where n is the total number of tests.\",\"breadcrumb\":{\"@id\":\"https:\/\/www.uni.lu\/fstm-fr\/events\/false-discovery-rate-control-with-unknown-null-distribution\/#breadcrumb\"},\"inLanguage\":\"fr-FR\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/www.uni.lu\/fstm-fr\/events\/false-discovery-rate-control-with-unknown-null-distribution\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"fr-FR\",\"@id\":\"https:\/\/www.uni.lu\/fstm-fr\/events\/false-discovery-rate-control-with-unknown-null-distribution\/#primaryimage\",\"url\":\"https:\/\/www.uni.lu\/wp-content\/uploads\/sites\/20\/2021\/03\/default-4.jpg\",\"contentUrl\":\"https:\/\/www.uni.lu\/wp-content\/uploads\/sites\/20\/2021\/03\/default-4.jpg\",\"width\":1500,\"height\":1125},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/www.uni.lu\/fstm-fr\/events\/false-discovery-rate-control-with-unknown-null-distribution\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\/\/www.uni.lu\/fr\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"Facult\u00e9 des Sciences, des Technologies et de M\u00e9decine\",\"item\":\"https:\/\/www.uni.lu\/fstm-fr\/\"},{\"@type\":\"ListItem\",\"position\":3,\"name\":\"Events\",\"item\":\"https:\/\/www.uni.lu\/fstm-fr\/events\/\"},{\"@type\":\"ListItem\",\"position\":4,\"name\":\"False Discovery rate control with unknown null distribution\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/www.uni.lu\/fstm-fr\/#website\",\"url\":\"https:\/\/www.uni.lu\/fstm-fr\/\",\"name\":\"FSTM\",\"description\":\"Facult\u00e9 des Sciences, des Technologies et de M\u00e9decine I Uni.lu\",\"publisher\":{\"@id\":\"https:\/\/www.uni.lu\/fstm-fr\/#organization\"},\"alternateName\":\"Facult\u00e9 des Sciences, des Technologies et de M\u00e9decine I Universit\u00e9 du Luxembourg\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/www.uni.lu\/fstm-fr\/?s={search_term_string}\"},\"query-input\":\"required name=search_term_string\"}],\"inLanguage\":\"fr-FR\"},{\"@type\":\"Organization\",\"@id\":\"https:\/\/www.uni.lu\/fstm-fr\/#organization\",\"name\":\"FSTM - Universit\u00e9 du Luxembourg I Uni.lu\",\"alternateName\":\"Facult\u00e9 des Sciences, des Technologies et de M\u00e9decine\",\"url\":\"https:\/\/www.uni.lu\/fstm-fr\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"fr-FR\",\"@id\":\"https:\/\/www.uni.lu\/fstm-fr\/#\/schema\/logo\/image\/\",\"url\":\"https:\/\/www.uni.lu\/wp-content\/uploads\/sites\/20\/2026\/03\/03111744\/FSTM_SM-Profile_1600x1600px-scaled.jpg\",\"contentUrl\":\"https:\/\/www.uni.lu\/wp-content\/uploads\/sites\/20\/2026\/03\/03111744\/FSTM_SM-Profile_1600x1600px-scaled.jpg\",\"width\":2560,\"height\":2560,\"caption\":\"FSTM - Universit\u00e9 du Luxembourg I Uni.lu\"},\"image\":{\"@id\":\"https:\/\/www.uni.lu\/fstm-fr\/#\/schema\/logo\/image\/\"},\"sameAs\":[\"https:\/\/www.facebook.com\/fstm.uni.lu\/\",\"https:\/\/www.linkedin.com\/showcase\/fstm-uni-lu\"]}]}<\/script>\n<!-- \/ Yoast SEO Premium plugin. -->","yoast_head_json":{"title":"False Discovery rate control with unknown null distribution - FSTM I Uni.lu","description":"Classical multiple testing theory prescribes the null distribution, which is often a too stringent assumption for nowadays large scale experiments. This paper presents theoretical foundations to understand the limitations caused by ignoring the null distribution, and how it can be properly learned from the (same) data-set, when possible. While an oracle procedure in that case is the Benjamini Hochberg procedure applied with the true (unknown) null distribution, we pursue the aim of building a procedure that asymptotically mimics the performance of the oracle (AMO in short). For a Gaussian null, our main result states that an AMO procedure exists if and only if the sparsity parameter k (number of false nulls) is of order less than n\/log(n), where n is the total number of tests.","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/www.uni.lu\/fstm-fr\/events\/false-discovery-rate-control-with-unknown-null-distribution\/","og_locale":"fr_FR","og_type":"article","og_title":"False Discovery rate control with unknown null distribution","og_description":"Classical multiple testing theory prescribes the null distribution, which is often a too stringent assumption for nowadays large scale experiments. This paper presents theoretical foundations to understand the limitations caused by ignoring the null distribution, and how it can be properly learned from the (same) data-set, when possible. While an oracle procedure in that case is the Benjamini Hochberg procedure applied with the true (unknown) null distribution, we pursue the aim of building a procedure that asymptotically mimics the performance of the oracle (AMO in short). For a Gaussian null, our main result states that an AMO procedure exists if and only if the sparsity parameter k (number of false nulls) is of order less than n\/log(n), where n is the total number of tests.","og_url":"https:\/\/www.uni.lu\/fstm-fr\/events\/false-discovery-rate-control-with-unknown-null-distribution\/","og_site_name":"FSTM FR","article_publisher":"https:\/\/www.facebook.com\/fstm.uni.lu\/","og_image":[{"width":2560,"height":2560,"url":"https:\/\/www.uni.lu\/wp-content\/uploads\/sites\/20\/2026\/03\/03111744\/FSTM_SM-Profile_1600x1600px-scaled.jpg","type":"image\/jpeg"}],"twitter_card":"summary_large_image","twitter_misc":{"Dur\u00e9e de lecture estim\u00e9e":"1 minute"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/www.uni.lu\/fstm-fr\/events\/false-discovery-rate-control-with-unknown-null-distribution\/","url":"https:\/\/www.uni.lu\/fstm-fr\/events\/false-discovery-rate-control-with-unknown-null-distribution\/","name":"False Discovery rate control with unknown null distribution - FSTM I Uni.lu","isPartOf":{"@id":"https:\/\/www.uni.lu\/fstm-fr\/#website"},"primaryImageOfPage":{"@id":"https:\/\/www.uni.lu\/fstm-fr\/events\/false-discovery-rate-control-with-unknown-null-distribution\/#primaryimage"},"image":{"@id":"https:\/\/www.uni.lu\/fstm-fr\/events\/false-discovery-rate-control-with-unknown-null-distribution\/#primaryimage"},"thumbnailUrl":"https:\/\/www.uni.lu\/wp-content\/uploads\/sites\/20\/2021\/03\/default-4.jpg","datePublished":"2021-03-16T07:22:29+00:00","dateModified":"2021-03-16T07:22:29+00:00","description":"Classical multiple testing theory prescribes the null distribution, which is often a too stringent assumption for nowadays large scale experiments. This paper presents theoretical foundations to understand the limitations caused by ignoring the null distribution, and how it can be properly learned from the (same) data-set, when possible. While an oracle procedure in that case is the Benjamini Hochberg procedure applied with the true (unknown) null distribution, we pursue the aim of building a procedure that asymptotically mimics the performance of the oracle (AMO in short). For a Gaussian null, our main result states that an AMO procedure exists if and only if the sparsity parameter k (number of false nulls) is of order less than n\/log(n), where n is the total number of tests.","breadcrumb":{"@id":"https:\/\/www.uni.lu\/fstm-fr\/events\/false-discovery-rate-control-with-unknown-null-distribution\/#breadcrumb"},"inLanguage":"fr-FR","potentialAction":[{"@type":"ReadAction","target":["https:\/\/www.uni.lu\/fstm-fr\/events\/false-discovery-rate-control-with-unknown-null-distribution\/"]}]},{"@type":"ImageObject","inLanguage":"fr-FR","@id":"https:\/\/www.uni.lu\/fstm-fr\/events\/false-discovery-rate-control-with-unknown-null-distribution\/#primaryimage","url":"https:\/\/www.uni.lu\/wp-content\/uploads\/sites\/20\/2021\/03\/default-4.jpg","contentUrl":"https:\/\/www.uni.lu\/wp-content\/uploads\/sites\/20\/2021\/03\/default-4.jpg","width":1500,"height":1125},{"@type":"BreadcrumbList","@id":"https:\/\/www.uni.lu\/fstm-fr\/events\/false-discovery-rate-control-with-unknown-null-distribution\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/www.uni.lu\/fr"},{"@type":"ListItem","position":2,"name":"Facult\u00e9 des Sciences, des Technologies et de M\u00e9decine","item":"https:\/\/www.uni.lu\/fstm-fr\/"},{"@type":"ListItem","position":3,"name":"Events","item":"https:\/\/www.uni.lu\/fstm-fr\/events\/"},{"@type":"ListItem","position":4,"name":"False Discovery rate control with unknown null distribution"}]},{"@type":"WebSite","@id":"https:\/\/www.uni.lu\/fstm-fr\/#website","url":"https:\/\/www.uni.lu\/fstm-fr\/","name":"FSTM","description":"Facult\u00e9 des Sciences, des Technologies et de M\u00e9decine I Uni.lu","publisher":{"@id":"https:\/\/www.uni.lu\/fstm-fr\/#organization"},"alternateName":"Facult\u00e9 des Sciences, des Technologies et de M\u00e9decine I Universit\u00e9 du Luxembourg","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/www.uni.lu\/fstm-fr\/?s={search_term_string}"},"query-input":"required name=search_term_string"}],"inLanguage":"fr-FR"},{"@type":"Organization","@id":"https:\/\/www.uni.lu\/fstm-fr\/#organization","name":"FSTM - Universit\u00e9 du Luxembourg I Uni.lu","alternateName":"Facult\u00e9 des Sciences, des Technologies et de M\u00e9decine","url":"https:\/\/www.uni.lu\/fstm-fr\/","logo":{"@type":"ImageObject","inLanguage":"fr-FR","@id":"https:\/\/www.uni.lu\/fstm-fr\/#\/schema\/logo\/image\/","url":"https:\/\/www.uni.lu\/wp-content\/uploads\/sites\/20\/2026\/03\/03111744\/FSTM_SM-Profile_1600x1600px-scaled.jpg","contentUrl":"https:\/\/www.uni.lu\/wp-content\/uploads\/sites\/20\/2026\/03\/03111744\/FSTM_SM-Profile_1600x1600px-scaled.jpg","width":2560,"height":2560,"caption":"FSTM - Universit\u00e9 du Luxembourg I Uni.lu"},"image":{"@id":"https:\/\/www.uni.lu\/fstm-fr\/#\/schema\/logo\/image\/"},"sameAs":["https:\/\/www.facebook.com\/fstm.uni.lu\/","https:\/\/www.linkedin.com\/showcase\/fstm-uni-lu"]}]}},"_links":{"self":[{"href":"https:\/\/www.uni.lu\/fstm-fr\/wp-json\/wp\/v2\/events\/1731"}],"collection":[{"href":"https:\/\/www.uni.lu\/fstm-fr\/wp-json\/wp\/v2\/events"}],"about":[{"href":"https:\/\/www.uni.lu\/fstm-fr\/wp-json\/wp\/v2\/types\/events"}],"replies":[{"embeddable":true,"href":"https:\/\/www.uni.lu\/fstm-fr\/wp-json\/wp\/v2\/comments?post=1731"}],"version-history":[{"count":0,"href":"https:\/\/www.uni.lu\/fstm-fr\/wp-json\/wp\/v2\/events\/1731\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.uni.lu\/fstm-fr\/wp-json\/wp\/v2\/media\/1732"}],"wp:attachment":[{"href":"https:\/\/www.uni.lu\/fstm-fr\/wp-json\/wp\/v2\/media?parent=1731"}],"wp:term":[{"taxonomy":"events-topic","embeddable":true,"href":"https:\/\/www.uni.lu\/fstm-fr\/wp-json\/wp\/v2\/events-topic?post=1731"},{"taxonomy":"events-type","embeddable":true,"href":"https:\/\/www.uni.lu\/fstm-fr\/wp-json\/wp\/v2\/events-type?post=1731"},{"taxonomy":"organisation","embeddable":true,"href":"https:\/\/www.uni.lu\/fstm-fr\/wp-json\/wp\/v2\/organisation?post=1731"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}