{"id":1133,"date":"2024-02-22T19:24:49","date_gmt":"2024-02-22T19:24:49","guid":{"rendered":"https:\/\/prlalmeida.com.br\/dsbd\/?page_id=1133"},"modified":"2024-11-27T00:47:46","modified_gmt":"2024-11-27T00:47:46","slug":"concept-drifts","status":"publish","type":"page","link":"https:\/\/prlalmeida.com.br\/dsbd\/concept-drifts\/","title":{"rendered":"Concept Drifts"},"content":{"rendered":"\t\t<div data-elementor-type=\"wp-page\" data-elementor-id=\"1133\" class=\"elementor elementor-1133\">\n\t\t\t\t<div class=\"elementor-element elementor-element-989a78d e-flex e-con-boxed e-con e-parent\" data-id=\"989a78d\" data-element_type=\"container\" data-e-type=\"container\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-5e6d569 elementor-widget elementor-widget-heading\" data-id=\"5e6d569\" data-element_type=\"widget\" data-e-type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<h2 class=\"elementor-heading-title elementor-size-default\">Concept Drifts <\/h2>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t<div class=\"elementor-element elementor-element-db71cc1 e-flex e-con-boxed e-con e-parent\" data-id=\"db71cc1\" data-element_type=\"container\" data-e-type=\"container\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-3694153 elementor-widget elementor-widget-text-editor\" data-id=\"3694153\" data-element_type=\"widget\" data-e-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\t<p><span style=\"font-weight: 400;\">Under non-stationary environments, machine learning models must be constantly updated, and changes must be detected to make the models compatible with the current environment conditions. These changes that force the model&#8217;s updates are known as Concept Drifts. A model that cannot cope with such changes will have its performance degraded over time.<\/span><\/p><p><span style=\"font-weight: 400;\">For example, a model must be constantly updated when dealing with fraud detection since attackers are constantly creating new fraud approaches.<\/span><\/p><p><span style=\"font-weight: 400;\">In these scenarios, both model updates and predictions must be made in real-time due to time constraints and the fast arrival of samples.<\/span><\/p>\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\t\t<\/div>\n\t\t<div class=\"elementor-element elementor-element-18ace9b e-flex e-con-boxed e-con e-parent\" data-id=\"18ace9b\" data-element_type=\"container\" data-e-type=\"container\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-f4e9876 elementor-widget elementor-widget-heading\" data-id=\"f4e9876\" data-element_type=\"widget\" data-e-type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<h2 class=\"elementor-heading-title elementor-size-default\">\nCurrent Results\n<\/h2>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t<div class=\"elementor-element elementor-element-cad56d1 e-flex e-con-boxed e-con e-parent\" data-id=\"cad56d1\" data-element_type=\"container\" data-e-type=\"container\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-261b808 elementor-widget elementor-widget-heading\" data-id=\"261b808\" data-element_type=\"widget\" data-e-type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<h2 class=\"elementor-heading-title elementor-size-default\">The Dynse Framework.<\/h2>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t<div class=\"elementor-element elementor-element-e804c56 e-flex e-con-boxed e-con e-parent\" data-id=\"e804c56\" data-element_type=\"container\" data-e-type=\"container\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-45dde06 elementor-widget elementor-widget-text-editor\" data-id=\"45dde06\" data-element_type=\"widget\" data-e-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\t<p><span style=\"font-weight: 400;\">The Dynse Framework is able to adapt dynamic selection methods to deal with non-stationary environments containing concept drifts. It achieves state-of-the-art results in well-known benchmarks, and it is publicly available at https:\/\/github.com\/paulorla\/dynse<\/span><\/p>\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\t\t<\/div>\n\t\t<div class=\"elementor-element elementor-element-11a00ff e-flex e-con-boxed e-con e-parent\" data-id=\"11a00ff\" data-element_type=\"container\" data-e-type=\"container\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-f8d154c elementor-widget__width-initial elementor-widget elementor-widget-image\" data-id=\"f8d154c\" data-element_type=\"widget\" data-e-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\t\t\t<img fetchpriority=\"high\" decoding=\"async\" width=\"818\" height=\"679\" src=\"https:\/\/prlalmeida.com.br\/dsbd\/wp-content\/uploads\/2024\/02\/Screenshot-2023-12-21-162603.png\" class=\"attachment-large size-large wp-image-1136\" alt=\"\" srcset=\"https:\/\/prlalmeida.com.br\/dsbd\/wp-content\/uploads\/2024\/02\/Screenshot-2023-12-21-162603.png 818w, https:\/\/prlalmeida.com.br\/dsbd\/wp-content\/uploads\/2024\/02\/Screenshot-2023-12-21-162603-300x249.png 300w, https:\/\/prlalmeida.com.br\/dsbd\/wp-content\/uploads\/2024\/02\/Screenshot-2023-12-21-162603-768x637.png 768w\" sizes=\"(max-width: 818px) 100vw, 818px\" \/>\t\t\t\t\t\t\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\t\t<\/div>\n\t\t<div class=\"elementor-element elementor-element-5bad938 e-flex e-con-boxed e-con e-parent\" data-id=\"5bad938\" data-element_type=\"container\" data-e-type=\"container\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-9eefb75 elementor-widget elementor-widget-heading\" data-id=\"9eefb75\" data-element_type=\"widget\" data-e-type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<h2 class=\"elementor-heading-title elementor-size-default\">Publications about the Dynse<\/h2>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t<div class=\"elementor-element elementor-element-58047f8 e-flex e-con-boxed e-con e-parent\" data-id=\"58047f8\" data-element_type=\"container\" data-e-type=\"container\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-0fc5d61 elementor-widget elementor-widget-text-editor\" data-id=\"0fc5d61\" data-element_type=\"widget\" data-e-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\t<p><a class=\"projects-title\" style=\"text-decoration: none;\" href=\"https:\/\/www.sciencedirect.com\/science\/article\/pii\/S0957417418301611\">Almeida et Al. Adapting Dynamic Classifier Selection for Concept Drift. 2018.<\/a><\/p>\n\n<a class=\"projects-title\" style=\"text-decoration: none;\" href=\"https:\/\/ieeexplore.ieee.org\/abstract\/document\/7814713\">Almeida et Al. Handling Concept Drifts Using Dynamic Selection of Classifiers. 2016<\/a>\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\t\t<\/div>\n\t\t<div class=\"elementor-element elementor-element-37b0549 e-flex e-con-boxed e-con e-parent\" data-id=\"37b0549\" data-element_type=\"container\" data-e-type=\"container\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-fff80d0 elementor-widget elementor-widget-heading\" data-id=\"fff80d0\" data-element_type=\"widget\" data-e-type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<h2 class=\"elementor-heading-title elementor-size-default\">Distance Function and Normalization Under Stream Scenarios<\/h2>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t<div class=\"elementor-element elementor-element-bee8c8e e-flex e-con-boxed e-con e-parent\" data-id=\"bee8c8e\" data-element_type=\"container\" data-e-type=\"container\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-80a0770 elementor-widget elementor-widget-text-editor\" data-id=\"80a0770\" data-element_type=\"widget\" data-e-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\t<p><span style=\"font-weight: 400;\">Measuring the distance when no normalization is possible is a challenge since we do not know the exact range of the feature space. For data streams, if we normalize the data, we should consider strategies that we can reply in real environments, such as using the statistics of a recent batch, or to use a sample of the dataset. Furthermore, maybe the normalization is not mandatory for data streams. Giving information to our Machine Learning models that we are not sure of \u2014 i.e., the range of the data \u2014 is not the best approach.<\/span><\/p><p><span style=\"font-weight: 400;\">One example is the temperature, which varies a lot throughout the year. Using the statistics of the summer to normalize data from the winter is not valid. In the paper <\/span><i><span style=\"font-weight: 400;\">Distance Function and Normalization Under Stream Scenarios,<\/span><\/i><span style=\"font-weight: 400;\"> we conclude that the non-normalization of data leads is a good strategy under concept drifting streams.<\/span><\/p><p><span style=\"font-weight: 400;\">We also conclude that the Canberra and the Manhattan Distance functions were the best performing ones when the range of the features varied with time.<\/span><\/p>\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\t\t<\/div>\n\t\t<div class=\"elementor-element elementor-element-e91716e e-flex e-con-boxed e-con e-parent\" data-id=\"e91716e\" data-element_type=\"container\" data-e-type=\"container\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-19f7e83 elementor-widget__width-initial elementor-widget elementor-widget-image\" data-id=\"19f7e83\" data-element_type=\"widget\" data-e-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\t\t\t<img decoding=\"async\" width=\"655\" height=\"421\" src=\"https:\/\/prlalmeida.com.br\/dsbd\/wp-content\/uploads\/2024\/02\/Screenshot-2023-12-21-162641.png\" class=\"attachment-large size-large wp-image-1144\" alt=\"\" srcset=\"https:\/\/prlalmeida.com.br\/dsbd\/wp-content\/uploads\/2024\/02\/Screenshot-2023-12-21-162641.png 655w, https:\/\/prlalmeida.com.br\/dsbd\/wp-content\/uploads\/2024\/02\/Screenshot-2023-12-21-162641-300x193.png 300w\" sizes=\"(max-width: 655px) 100vw, 655px\" \/>\t\t\t\t\t\t\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\t\t<\/div>\n\t\t<div class=\"elementor-element elementor-element-e7882a5 e-flex e-con-boxed e-con e-parent\" data-id=\"e7882a5\" data-element_type=\"container\" data-e-type=\"container\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-dd69b3c elementor-widget elementor-widget-heading\" data-id=\"dd69b3c\" data-element_type=\"widget\" data-e-type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<h2 class=\"elementor-heading-title elementor-size-default\">Publications about the Dynse<\/h2>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t<div class=\"elementor-element elementor-element-29170af e-flex e-con-boxed e-con e-parent\" data-id=\"29170af\" data-element_type=\"container\" data-e-type=\"container\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-e86264e elementor-widget elementor-widget-text-editor\" data-id=\"e86264e\" data-element_type=\"widget\" data-e-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\t<a class=\"projects-title\" style=\"text-decoration: none;\" href=\"https:\/\/www.sciencedirect.com\/science\/article\/pii\/S0957417418301611\">Barboza et al. Distance Functions and Normalization Under Stream Scenarios.<\/a>\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\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t","protected":false},"excerpt":{"rendered":"<p>Concept Drifts Under non-stationary environments, machine learning models must be constantly updated, and changes must be detected to make the [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":0,"parent":0,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"","meta":{"site-sidebar-layout":"no-sidebar","site-content-layout":"","ast-site-content-layout":"full-width-container","site-content-style":"default","site-sidebar-style":"default","ast-global-header-display":"","ast-banner-title-visibility":"","ast-main-header-display":"","ast-hfb-above-header-display":"","ast-hfb-below-header-display":"","ast-hfb-mobile-header-display":"","site-post-title":"disabled","ast-breadcrumbs-content":"","ast-featured-img":"disabled","footer-sml-layout":"","theme-transparent-header-meta":"","adv-header-id-meta":"","stick-header-meta":"","header-above-stick-meta":"","header-main-stick-meta":"","header-below-stick-meta":"","astra-migrate-meta-layouts":"default","ast-page-background-enabled":"default","ast-page-background-meta":{"desktop":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"ast-content-background-meta":{"desktop":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"footnotes":""},"class_list":["post-1133","page","type-page","status-publish","hentry"],"_links":{"self":[{"href":"https:\/\/prlalmeida.com.br\/dsbd\/wp-json\/wp\/v2\/pages\/1133","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/prlalmeida.com.br\/dsbd\/wp-json\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/prlalmeida.com.br\/dsbd\/wp-json\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/prlalmeida.com.br\/dsbd\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/prlalmeida.com.br\/dsbd\/wp-json\/wp\/v2\/comments?post=1133"}],"version-history":[{"count":19,"href":"https:\/\/prlalmeida.com.br\/dsbd\/wp-json\/wp\/v2\/pages\/1133\/revisions"}],"predecessor-version":[{"id":2064,"href":"https:\/\/prlalmeida.com.br\/dsbd\/wp-json\/wp\/v2\/pages\/1133\/revisions\/2064"}],"wp:attachment":[{"href":"https:\/\/prlalmeida.com.br\/dsbd\/wp-json\/wp\/v2\/media?parent=1133"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}