{"id":31351,"date":"2023-02-21T16:31:51","date_gmt":"2023-02-21T08:31:51","guid":{"rendered":"https:\/\/www.hkmu.edu.hk\/st\/?p=31351"},"modified":"2025-01-16T15:06:26","modified_gmt":"2025-01-16T07:06:26","slug":"crant-talks-series-1","status":"publish","type":"post","link":"https:\/\/www.hkmu.edu.hk\/st\/events\/crant-talks-series-1\/","title":{"rendered":"CRANT Talk Series: FedCorr: Multi-Stage Federated Learning for Label Noise Correction"},"content":{"rendered":"\t\t<div data-elementor-type=\"wp-page\" data-elementor-id=\"31351\" class=\"elementor elementor-31351\" data-elementor-settings=\"[]\">\n\t\t\t\t\t\t\t<div class=\"elementor-section-wrap\">\n\t\t\t\t\t\t\t<section class=\"has_eae_slider wavo-column-gap-default elementor-section elementor-top-section elementor-element elementor-element-6e6b43b elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"6e6b43b\" data-element_type=\"section\" data-settings=\"{&quot;jet_parallax_layout_list&quot;:[{&quot;jet_parallax_layout_image&quot;:{&quot;url&quot;:&quot;&quot;,&quot;id&quot;:&quot;&quot;},&quot;_id&quot;:&quot;fc47f36&quot;,&quot;jet_parallax_layout_image_tablet&quot;:{&quot;url&quot;:&quot;&quot;,&quot;id&quot;:&quot;&quot;},&quot;jet_parallax_layout_image_mobile&quot;:{&quot;url&quot;:&quot;&quot;,&quot;id&quot;:&quot;&quot;},&quot;jet_parallax_layout_speed&quot;:{&quot;unit&quot;:&quot;%&quot;,&quot;size&quot;:50,&quot;sizes&quot;:[]},&quot;jet_parallax_layout_type&quot;:&quot;scroll&quot;,&quot;jet_parallax_layout_direction&quot;:null,&quot;jet_parallax_layout_fx_direction&quot;:null,&quot;jet_parallax_layout_z_index&quot;:&quot;&quot;,&quot;jet_parallax_layout_bg_x&quot;:50,&quot;jet_parallax_layout_bg_x_tablet&quot;:&quot;&quot;,&quot;jet_parallax_layout_bg_x_mobile&quot;:&quot;&quot;,&quot;jet_parallax_layout_bg_y&quot;:50,&quot;jet_parallax_layout_bg_y_tablet&quot;:&quot;&quot;,&quot;jet_parallax_layout_bg_y_mobile&quot;:&quot;&quot;,&quot;jet_parallax_layout_bg_size&quot;:&quot;auto&quot;,&quot;jet_parallax_layout_bg_size_tablet&quot;:&quot;&quot;,&quot;jet_parallax_layout_bg_size_mobile&quot;:&quot;&quot;,&quot;jet_parallax_layout_animation_prop&quot;:&quot;transform&quot;,&quot;jet_parallax_layout_on&quot;:[&quot;desktop&quot;,&quot;tablet&quot;]}]}\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"has_eae_slider elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-ccf5962\" data-id=\"ccf5962\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-61610e5 elementor-widget elementor-widget-premium-addon-title\" data-id=\"61610e5\" data-element_type=\"widget\" data-widget_type=\"premium-addon-title.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\n\t<div class=\"premium-title-container style1\">\n\t\t<h6 class=\"premium-title-header premium-title-style1\">\n\t\t\t\n\t\t\t\n\t\t\t\t\t\t\t\t\t<span class=\"premium-title-text\" >\n\t\t\t\tDate: Friday 24 February 2023 <br>Time: 2:00 PM \u2013 3:00 PM<br>Location: D1115, JCC, HKMU\t\t\t<\/span>\n\t\t\t\t\n\t\t\t\t\t\t\t\t<\/h6>\n\t<\/div>\n\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"has_eae_slider wavo-column-gap-default elementor-section elementor-top-section elementor-element elementor-element-e2e6754 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"e2e6754\" data-element_type=\"section\" data-settings=\"{&quot;jet_parallax_layout_list&quot;:[{&quot;jet_parallax_layout_image&quot;:{&quot;url&quot;:&quot;&quot;,&quot;id&quot;:&quot;&quot;},&quot;_id&quot;:&quot;de876f7&quot;,&quot;jet_parallax_layout_image_tablet&quot;:{&quot;url&quot;:&quot;&quot;,&quot;id&quot;:&quot;&quot;},&quot;jet_parallax_layout_image_mobile&quot;:{&quot;url&quot;:&quot;&quot;,&quot;id&quot;:&quot;&quot;},&quot;jet_parallax_layout_speed&quot;:{&quot;unit&quot;:&quot;%&quot;,&quot;size&quot;:50,&quot;sizes&quot;:[]},&quot;jet_parallax_layout_type&quot;:&quot;scroll&quot;,&quot;jet_parallax_layout_direction&quot;:null,&quot;jet_parallax_layout_fx_direction&quot;:null,&quot;jet_parallax_layout_z_index&quot;:&quot;&quot;,&quot;jet_parallax_layout_bg_x&quot;:50,&quot;jet_parallax_layout_bg_x_tablet&quot;:&quot;&quot;,&quot;jet_parallax_layout_bg_x_mobile&quot;:&quot;&quot;,&quot;jet_parallax_layout_bg_y&quot;:50,&quot;jet_parallax_layout_bg_y_tablet&quot;:&quot;&quot;,&quot;jet_parallax_layout_bg_y_mobile&quot;:&quot;&quot;,&quot;jet_parallax_layout_bg_size&quot;:&quot;auto&quot;,&quot;jet_parallax_layout_bg_size_tablet&quot;:&quot;&quot;,&quot;jet_parallax_layout_bg_size_mobile&quot;:&quot;&quot;,&quot;jet_parallax_layout_animation_prop&quot;:&quot;transform&quot;,&quot;jet_parallax_layout_on&quot;:[&quot;desktop&quot;,&quot;tablet&quot;]}]}\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"has_eae_slider elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-e703f28\" data-id=\"e703f28\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-2f97b9c elementor-widget elementor-widget-text-editor\" data-id=\"2f97b9c\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t<p><strong><em>Title<\/em><\/strong>:\u00a0FedCorr: Multi-Stage Federated Learning for Label Noise Correction<\/p><p><strong><em>Abstract<\/em><\/strong>:\u00a0Federated learning (FL) is a privacy-preserving distributed learning paradigm that enables clients to jointly train a global model. In real-world FL implementations, client data could have label noise, and different clients could have vastly different label noise levels. Although there exist methods in centralized learning for tackling label noise, such methods do not perform well on heterogeneous label noise in FL settings, due to the typically smaller sizes of client datasets and data privacy requirements in FL. In this talk, we propose FedCorr, a general multi-stage framework to tackle heterogeneous label noise in FL, without making any assumptions on the noise models of local clients, while still maintaining client data privacy. In particular, (1) FedCorr dynamically identifies noisy clients by exploiting the dimensionalities of the model prediction subspaces independently measured on all clients, and then identifies incorrect labels on noisy clients based on per-sample losses. To deal with data heterogeneity and to increase training stability, we propose an adaptive local proximal regularization term that is based on estimated local noise levels. (2) We further finetune the global model on identified clean clients and correct the noisy labels for the remaining noisy clients after finetuning. (3) Finally, we apply the usual training on all clients to make full use of all local data. Experiments conducted on CIFAR-10\/100 with federated synthetic label noise, and on a real-world noisy dataset, Clothing1M, demonstrate that FedCorr is robust to label noise and substantially outperforms the state-of-the-art methods at multiple noise levels.<\/p><p><strong><em>Bio of Speaker<\/em><\/strong>:\u00a0Jingyi Xu is a Ph.D. Candidate at Singapore University of Technology and Design, under the supervision of Prof. Ernest Chong and Prof. Tony Quek. Before joining SUTD, she received her Bachelor&#8217;s degree from Fudan University. Her research interests lie in algebraic machine reasoning, federated learning, and robust learning with label noise.<\/p><p><strong><em>Language:<\/em><\/strong>\u00a0English<\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"has_eae_slider wavo-column-gap-default elementor-section elementor-top-section elementor-element elementor-element-2fa34d1 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"2fa34d1\" data-element_type=\"section\" data-settings=\"{&quot;jet_parallax_layout_list&quot;:[{&quot;jet_parallax_layout_image&quot;:{&quot;url&quot;:&quot;&quot;,&quot;id&quot;:&quot;&quot;},&quot;_id&quot;:&quot;4dd5f14&quot;,&quot;jet_parallax_layout_image_tablet&quot;:{&quot;url&quot;:&quot;&quot;,&quot;id&quot;:&quot;&quot;},&quot;jet_parallax_layout_image_mobile&quot;:{&quot;url&quot;:&quot;&quot;,&quot;id&quot;:&quot;&quot;},&quot;jet_parallax_layout_speed&quot;:{&quot;unit&quot;:&quot;%&quot;,&quot;size&quot;:50,&quot;sizes&quot;:[]},&quot;jet_parallax_layout_type&quot;:&quot;scroll&quot;,&quot;jet_parallax_layout_direction&quot;:null,&quot;jet_parallax_layout_fx_direction&quot;:null,&quot;jet_parallax_layout_z_index&quot;:&quot;&quot;,&quot;jet_parallax_layout_bg_x&quot;:50,&quot;jet_parallax_layout_bg_x_tablet&quot;:&quot;&quot;,&quot;jet_parallax_layout_bg_x_mobile&quot;:&quot;&quot;,&quot;jet_parallax_layout_bg_y&quot;:50,&quot;jet_parallax_layout_bg_y_tablet&quot;:&quot;&quot;,&quot;jet_parallax_layout_bg_y_mobile&quot;:&quot;&quot;,&quot;jet_parallax_layout_bg_size&quot;:&quot;auto&quot;,&quot;jet_parallax_layout_bg_size_tablet&quot;:&quot;&quot;,&quot;jet_parallax_layout_bg_size_mobile&quot;:&quot;&quot;,&quot;jet_parallax_layout_animation_prop&quot;:&quot;transform&quot;,&quot;jet_parallax_layout_on&quot;:[&quot;desktop&quot;,&quot;tablet&quot;]}]}\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"has_eae_slider elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-a5463cc\" data-id=\"a5463cc\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-52e5dd9 elementor-widget elementor-widget-image\" data-id=\"52e5dd9\" data-element_type=\"widget\" data-widget_type=\"image.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t\t\t\t<figure class=\"wp-caption\">\n\t\t\t\t\t\t\t\t\t\t<img width=\"960\" height=\"720\" src=\"https:\/\/www.hkmu.edu.hk\/st\/wp-content\/uploads\/sites\/25\/2023\/02\/CRANT-Seminar-1-1024x768.jpg\" class=\"attachment-large size-large\" alt=\"\" decoding=\"async\" loading=\"lazy\" srcset=\"https:\/\/www.hkmu.edu.hk\/st\/wp-content\/uploads\/sites\/25\/2023\/02\/CRANT-Seminar-1-1024x768.jpg 1024w, https:\/\/www.hkmu.edu.hk\/st\/wp-content\/uploads\/sites\/25\/2023\/02\/CRANT-Seminar-1-300x225.jpg 300w, https:\/\/www.hkmu.edu.hk\/st\/wp-content\/uploads\/sites\/25\/2023\/02\/CRANT-Seminar-1-768x576.jpg 768w, https:\/\/www.hkmu.edu.hk\/st\/wp-content\/uploads\/sites\/25\/2023\/02\/CRANT-Seminar-1-1536x1152.jpg 1536w, https:\/\/www.hkmu.edu.hk\/st\/wp-content\/uploads\/sites\/25\/2023\/02\/CRANT-Seminar-1-500x375.jpg 500w, https:\/\/www.hkmu.edu.hk\/st\/wp-content\/uploads\/sites\/25\/2023\/02\/CRANT-Seminar-1.jpg 2016w\" sizes=\"(max-width: 960px) 100vw, 960px\" \/>\t\t\t\t\t\t\t\t\t\t\t<figcaption class=\"widget-image-caption wp-caption-text\">Group  Photo<\/figcaption>\n\t\t\t\t\t\t\t\t\t\t<\/figure>\n\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t","protected":false},"excerpt":{"rendered":"<p>Date: Friday 24 February 2023 Time: 2:00 PM \u2013 3:00 PM Location: D1115, JCC, HKMU Title:\u00a0FedCorr: Multi-Stage Federated Learning for Label Noise Correction Abstract:\u00a0Federated learning (FL) is a privacy-preserving distributed learning paradigm that enables clients to jointly train a global model. In real-world FL implementations, client data could have label noise, and different clients could...<\/p>\n","protected":false},"author":140,"featured_media":25326,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_expiration-date-status":"","_expiration-date":0,"_expiration-date-type":"","_expiration-date-categories":[],"_expiration-date-options":[]},"categories":[44],"tags":[],"yoast_head":"<!-- This site is optimized with the Yoast SEO Premium plugin v17.3 (Yoast SEO v21.2) - https:\/\/yoast.com\/wordpress\/plugins\/seo\/ -->\n<title>CRANT Talk Series: FedCorr: Multi-Stage Federated Learning for Label Noise Correction - School of Science and Technology - Hong Kong Metropolitan University<\/title>\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.hkmu.edu.hk\/st\/events\/crant-talks-series-1\/\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"CRANT Talk Series: FedCorr: Multi-Stage Federated Learning for Label Noise Correction\" \/>\n<meta property=\"og:description\" content=\"Date: Friday 24 February 2023 Time: 2:00 PM \u2013 3:00 PM Location: D1115, JCC, HKMU Title:\u00a0FedCorr: Multi-Stage Federated Learning for Label Noise Correction\" \/>\n<meta property=\"og:url\" content=\"https:\/\/www.hkmu.edu.hk\/st\/events\/crant-talks-series-1\/\" \/>\n<meta property=\"og:site_name\" content=\"School of Science and Technology - Hong Kong Metropolitan University\" \/>\n<meta property=\"article:published_time\" content=\"2023-02-21T08:31:51+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2025-01-16T07:06:26+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/www.hkmu.edu.hk\/st\/wp-content\/uploads\/sites\/25\/2022\/07\/6c3153d217ba3b6b73ffa7e8a63955ad.jpg\" \/>\n\t<meta property=\"og:image:width\" content=\"450\" \/>\n\t<meta property=\"og:image:height\" content=\"450\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/jpeg\" \/>\n<meta name=\"author\" content=\"hychan\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"Written by\" \/>\n\t<meta name=\"twitter:data1\" content=\"hychan\" \/>\n\t<meta name=\"twitter:label2\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data2\" content=\"2 minutes\" \/>\n<!-- \/ Yoast SEO Premium plugin. -->","yoast_head_json":{"title":"CRANT Talk Series: FedCorr: Multi-Stage Federated Learning for Label Noise Correction - School of Science and Technology - Hong Kong Metropolitan University","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.hkmu.edu.hk\/st\/events\/crant-talks-series-1\/","og_locale":"en_US","og_type":"article","og_title":"CRANT Talk Series: FedCorr: Multi-Stage Federated Learning for Label Noise Correction","og_description":"Date: Friday 24 February 2023 Time: 2:00 PM \u2013 3:00 PM Location: D1115, JCC, HKMU Title:\u00a0FedCorr: Multi-Stage Federated Learning for Label Noise Correction","og_url":"https:\/\/www.hkmu.edu.hk\/st\/events\/crant-talks-series-1\/","og_site_name":"School of Science and Technology - Hong Kong Metropolitan University","article_published_time":"2023-02-21T08:31:51+00:00","article_modified_time":"2025-01-16T07:06:26+00:00","og_image":[{"width":450,"height":450,"url":"https:\/\/www.hkmu.edu.hk\/st\/wp-content\/uploads\/sites\/25\/2022\/07\/6c3153d217ba3b6b73ffa7e8a63955ad.jpg","type":"image\/jpeg"}],"author":"hychan","twitter_card":"summary_large_image","twitter_misc":{"Written by":"hychan","Est. reading time":"2 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/www.hkmu.edu.hk\/st\/events\/crant-talks-series-1\/","url":"https:\/\/www.hkmu.edu.hk\/st\/events\/crant-talks-series-1\/","name":"CRANT Talk Series: FedCorr: Multi-Stage Federated Learning for Label Noise Correction - School of Science and Technology - Hong Kong Metropolitan University","isPartOf":{"@id":"https:\/\/www.hkmu.edu.hk\/st\/#website"},"datePublished":"2023-02-21T08:31:51+00:00","dateModified":"2025-01-16T07:06:26+00:00","author":{"@id":"https:\/\/www.hkmu.edu.hk\/st\/#\/schema\/person\/2029eddbfd142e49923243fc3c06a8f5"},"breadcrumb":{"@id":"https:\/\/www.hkmu.edu.hk\/st\/events\/crant-talks-series-1\/#breadcrumb"},"inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/www.hkmu.edu.hk\/st\/events\/crant-talks-series-1\/"]}]},{"@type":"BreadcrumbList","@id":"https:\/\/www.hkmu.edu.hk\/st\/events\/crant-talks-series-1\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"School of Science and Technology","item":"\/st\/"},{"@type":"ListItem","position":2,"name":"CRANT Talk Series: FedCorr: Multi-Stage Federated Learning for Label Noise Correction"}]},{"@type":"WebSite","@id":"https:\/\/www.hkmu.edu.hk\/st\/#website","url":"https:\/\/www.hkmu.edu.hk\/st\/","name":"School of Science and Technology - Hong Kong Metropolitan University","description":"School of Science and Technology - Hong Kong Metropolitan University","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/www.hkmu.edu.hk\/st\/?s={search_term_string}"},"query-input":"required name=search_term_string"}],"inLanguage":"en-US"},{"@type":"Person","@id":"https:\/\/www.hkmu.edu.hk\/st\/#\/schema\/person\/2029eddbfd142e49923243fc3c06a8f5","name":"hychan","image":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/www.hkmu.edu.hk\/st\/#\/schema\/person\/image\/","url":"https:\/\/secure.gravatar.com\/avatar\/1472dd6b98af305c5b29c8470b537bea?s=96&d=mm&r=g","contentUrl":"https:\/\/secure.gravatar.com\/avatar\/1472dd6b98af305c5b29c8470b537bea?s=96&d=mm&r=g","caption":"hychan"},"url":"https:\/\/www.hkmu.edu.hk\/st\/author\/hychan\/"}]}},"_links":{"self":[{"href":"https:\/\/www.hkmu.edu.hk\/st\/wp-json\/wp\/v2\/posts\/31351"}],"collection":[{"href":"https:\/\/www.hkmu.edu.hk\/st\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.hkmu.edu.hk\/st\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.hkmu.edu.hk\/st\/wp-json\/wp\/v2\/users\/140"}],"replies":[{"embeddable":true,"href":"https:\/\/www.hkmu.edu.hk\/st\/wp-json\/wp\/v2\/comments?post=31351"}],"version-history":[{"count":13,"href":"https:\/\/www.hkmu.edu.hk\/st\/wp-json\/wp\/v2\/posts\/31351\/revisions"}],"predecessor-version":[{"id":43006,"href":"https:\/\/www.hkmu.edu.hk\/st\/wp-json\/wp\/v2\/posts\/31351\/revisions\/43006"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.hkmu.edu.hk\/st\/wp-json\/wp\/v2\/media\/25326"}],"wp:attachment":[{"href":"https:\/\/www.hkmu.edu.hk\/st\/wp-json\/wp\/v2\/media?parent=31351"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.hkmu.edu.hk\/st\/wp-json\/wp\/v2\/categories?post=31351"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.hkmu.edu.hk\/st\/wp-json\/wp\/v2\/tags?post=31351"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}