{"id":219590,"date":"2024-01-24T06:30:00","date_gmt":"2024-01-24T06:30:00","guid":{"rendered":"https:\/\/datascientest.com\/?p=219590"},"modified":"2024-01-19T12:09:54","modified_gmt":"2024-01-19T12:09:54","slug":"recherche-automatique-darchitecture-neuronale-tout-savoir","status":"publish","type":"post","link":"https:\/\/datascientest.com\/recherche-automatique-darchitecture-neuronale-tout-savoir","title":{"rendered":"Recherche automatique d&rsquo;architecture neuronale : Tout ce qu&rsquo;il faut savoir"},"content":{"rendered":"\t\t<div data-elementor-type=\"wp-post\" data-elementor-id=\"219590\" class=\"elementor elementor-219590\" data-elementor-post-type=\"post\">\n\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-9c6c37d e-flex e-con-boxed e-con\" data-id=\"9c6c37d\" data-element_type=\"container\" data-settings=\"{&quot;content_width&quot;:&quot;boxed&quot;}\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-a130617 elementor-widget elementor-widget-heading\" data-id=\"a130617\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<style>\/*! elementor - v3.15.0 - 02-08-2023 *\/\n.elementor-heading-title{padding:0;margin:0;line-height:1}.elementor-widget-heading .elementor-heading-title[class*=elementor-size-]>a{color:inherit;font-size:inherit;line-height:inherit}.elementor-widget-heading .elementor-heading-title.elementor-size-small{font-size:15px}.elementor-widget-heading .elementor-heading-title.elementor-size-medium{font-size:19px}.elementor-widget-heading .elementor-heading-title.elementor-size-large{font-size:29px}.elementor-widget-heading .elementor-heading-title.elementor-size-xl{font-size:39px}.elementor-widget-heading .elementor-heading-title.elementor-size-xxl{font-size:59px}<\/style><h2 class=\"elementor-heading-title elementor-size-default\">L\u2019utilisation des mod\u00e8les de Deep Learning devient de plus en plus n\u00e9cessaire pour les entreprises. Cependant, son impl\u00e9mentation repr\u00e9sente un d\u00e9fi en termes de cr\u00e9ation d\u2019un mod\u00e8le qui s\u2019adapte bien aux besoins tout en restant efficace et \u00e9conomique en ressources. D\u2019o\u00f9 le besoin de doter l\u2019IA d'une compr\u00e9hension plus profonde et d\u2019une autonomie accrue.<\/h2>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-4a56293 elementor-widget elementor-widget-text-editor\" data-id=\"4a56293\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<style>\/*! elementor - v3.15.0 - 02-08-2023 *\/\n.elementor-widget-text-editor.elementor-drop-cap-view-stacked .elementor-drop-cap{background-color:#69727d;color:#fff}.elementor-widget-text-editor.elementor-drop-cap-view-framed .elementor-drop-cap{color:#69727d;border:3px solid;background-color:transparent}.elementor-widget-text-editor:not(.elementor-drop-cap-view-default) .elementor-drop-cap{margin-top:8px}.elementor-widget-text-editor:not(.elementor-drop-cap-view-default) .elementor-drop-cap-letter{width:1em;height:1em}.elementor-widget-text-editor .elementor-drop-cap{float:left;text-align:center;line-height:1;font-size:50px}.elementor-widget-text-editor .elementor-drop-cap-letter{display:inline-block}<\/style>\t\t\t\t<p><span style=\"font-weight: 400;\">Cet article vous explique tout sur la r\u00e9volution \u00e9mergente permettant de lib\u00e9rer les d\u00e9veloppeurs des t\u00e2ches fastidieuses du design et de l\u2019optimisation manuelle. L\u2019article met en \u00e9vidence les d\u00e9fis, le fonctionnement et les applications potentielles dans divers domaines laissant entrevoir un avenir fascinant pour l\u2019<strong>IA auto-apprenante<\/strong>.\u00a0<\/span><\/p>\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-54d5f57 e-flex e-con-boxed e-con\" data-id=\"54d5f57\" data-element_type=\"container\" data-settings=\"{&quot;content_width&quot;:&quot;boxed&quot;}\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-900d021 elementor-widget elementor-widget-heading\" data-id=\"900d021\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<h3 class=\"elementor-heading-title elementor-size-default\">D'o\u00f9 provient le LLM ?<\/h3>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-5bfd781 elementor-widget elementor-widget-text-editor\" data-id=\"5bfd781\" 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<p><span style=\"font-weight: 400;\">La recherche automatique d\u2019architecture neuronale (NAS, de l\u2019anglais <\/span><i><span style=\"font-weight: 400;\">Neural Architecture Search<\/span><\/i><span style=\"font-weight: 400;\">) est n\u00e9e du besoin d\u2019optimiser la cr\u00e9ation et le param\u00e9trage des<a href=\"https:\/\/datascientest.com\/machine-learning-tout-savoir\"> mod\u00e8les de Machine Learning<\/a>. La NAS essaye de r\u00e9pondre \u00e0 la question : est-ce possible de cr\u00e9er un algorithme capable de cr\u00e9er des r\u00e9seaux de neurones <\/span><b>mieux performants <\/b><span style=\"font-weight: 400;\">que ceux cr\u00e9\u00e9s et param\u00e9tr\u00e9s \u00e0 la main ? Les premiers articles portant ce sujet datent de 2016. Dans un premier temps <\/span><b>Barret Zoph<\/b><span style=\"font-weight: 400;\"> et <\/span><b>Quoc V<\/b><span style=\"font-weight: 400;\"> dans leur article <\/span><i><span style=\"font-weight: 400;\">Neural Architecture Search with Reinforcement Learning. <\/span><\/i><span style=\"font-weight: 400;\">Et peu apr\u00e8s, Bozen Baken, Otkrist Gupta, Nikhil Naik et Ramesh Raskar dans leur article <\/span><i><span style=\"font-weight: 400;\">Designing Neural Network Architectures using Reinforcement Learning.\u00a0<\/span><\/i><\/p><p><span style=\"font-weight: 400;\">En effet, les r\u00e9seaux de neurones sont souvent difficiles \u00e0 cr\u00e9er lorsque les probl\u00e8mes \u00e0 r\u00e9soudre deviennent plus complexes: la <\/span><b>cr\u00e9ation ne prend pas seulement beaucoup<\/b><span style=\"font-weight: 400;\"> de temps mais aussi de l\u2019effort, de la ma\u00eetrise sur le sujet et\u00a0 beaucoup d\u2019ajustements des param\u00e8tres. Donc, pour mieux optimiser ces ressources, les <\/span><b>NAS <\/b><span style=\"font-weight: 400;\">automatisent la cr\u00e9ation des <a href=\"https:\/\/datascientest.com\/fonctionnement-des-reseaux-neurones\">r\u00e9seaux de neurones<\/a>.<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-81e6ef8 elementor-align-center button-principale elementor-widget elementor-widget-button\" data-id=\"81e6ef8\" data-element_type=\"widget\" data-widget_type=\"button.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<div class=\"elementor-button-wrapper\">\n\t\t\t<a class=\"elementor-button elementor-button-link elementor-size-sm elementor-animation-grow\" href=\"https:\/\/datascientest.com\/formation-intelligence-artificielle\">\n\t\t\t\t\t\t<span class=\"elementor-button-content-wrapper\">\n\t\t\t\t\t\t<span class=\"elementor-button-text\">Se former en intelligence artificielle<\/span>\n\t\t<\/span>\n\t\t\t\t\t<\/a>\n\t\t<\/div>\n\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-1196614 e-flex e-con-boxed e-con\" data-id=\"1196614\" data-element_type=\"container\" data-settings=\"{&quot;content_width&quot;:&quot;boxed&quot;}\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-50ca79b elementor-widget elementor-widget-heading\" data-id=\"50ca79b\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<h3 class=\"elementor-heading-title elementor-size-default\">Comprendre la recherche automatique d\u2019architecture neuronale<\/h3>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-60fba56 elementor-widget elementor-widget-text-editor\" data-id=\"60fba56\" 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<p><span style=\"font-weight: 400;\">D\u00e8s sa conception, la recherche automatique d\u2019architecture neuronale constitue un sous-domaine de l\u2019<a href=\"https:\/\/datascientest.com\/automl-automatisation-menace-data-scientists\">AutoML<\/a> ou Machine Learning automatis\u00e9; <\/span><b>domaine charg\u00e9 d\u2019automatiser les t\u00e2ches<\/b><span style=\"font-weight: 400;\"> fastidieuses et it\u00e9ratives du d\u00e9veloppement de mod\u00e8les de Machine Learning. De sa part, l\u2019optimisation des <a href=\"https:\/\/datascientest.com\/hyperparametres-tout-savoir\">hyperparam\u00e8tres <\/a>est un sous-domaine des NAS.<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-8132603 elementor-widget elementor-widget-image\" data-id=\"8132603\" data-element_type=\"widget\" data-widget_type=\"image.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<style>\/*! elementor - v3.15.0 - 02-08-2023 *\/\n.elementor-widget-image{text-align:center}.elementor-widget-image a{display:inline-block}.elementor-widget-image a img[src$=\".svg\"]{width:48px}.elementor-widget-image img{vertical-align:middle;display:inline-block}<\/style>\t\t\t\t\t\t\t\t\t\t\t\t<img decoding=\"async\" width=\"600\" height=\"299\" src=\"https:\/\/datascientest.com\/wp-content\/uploads\/2023\/10\/recherche-automatique-d-architecture-neuronale1.jpg\" class=\"attachment-full size-full wp-image-219601\" alt=\"\" srcset=\"https:\/\/datascientest.com\/wp-content\/uploads\/2023\/10\/recherche-automatique-d-architecture-neuronale1.jpg 600w, https:\/\/datascientest.com\/wp-content\/uploads\/2023\/10\/recherche-automatique-d-architecture-neuronale1-300x150.jpg 300w\" sizes=\"(max-width: 600px) 100vw, 600px\" \/>\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<div class=\"elementor-element elementor-element-bff4af1 elementor-widget elementor-widget-text-editor\" data-id=\"bff4af1\" 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<p><span style=\"font-weight: 400;\">Une question se pose naturellement apr\u00e8s avoir d\u00e9fini le r\u00f4le des NAS : o\u00f9 participent-elles exactement dans le <\/span><b>processus du d\u00e9veloppement des mod\u00e8les<\/b><span style=\"font-weight: 400;\"> ? Et bien, la recherche automatique d\u2019architecture neuronale a lieu dans la partie train du flux de travail : comme son nom le dit bien, l\u2019on cherche l\u2019architecture optimale du <\/span><b>r\u00e9seau de neurones<\/b><span style=\"font-weight: 400;\">. <\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-1c08114 elementor-widget elementor-widget-image\" data-id=\"1c08114\" 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 decoding=\"async\" width=\"600\" height=\"300\" src=\"https:\/\/datascientest.com\/wp-content\/uploads\/2023\/10\/recherche-automatique-d-architecture-neuronale2.jpg\" class=\"attachment-full size-full wp-image-219602\" alt=\"\" srcset=\"https:\/\/datascientest.com\/wp-content\/uploads\/2023\/10\/recherche-automatique-d-architecture-neuronale2.jpg 600w, https:\/\/datascientest.com\/wp-content\/uploads\/2023\/10\/recherche-automatique-d-architecture-neuronale2-300x150.jpg 300w\" sizes=\"(max-width: 600px) 100vw, 600px\" \/>\t\t\t\t\t\t\t\t\t\t\t<figcaption class=\"widget-image-caption wp-caption-text\">Flux de travail<\/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\t\t<\/div>\n\t\t<div class=\"elementor-element elementor-element-1963a3d e-flex e-con-boxed e-con\" data-id=\"1963a3d\" data-element_type=\"container\" data-settings=\"{&quot;content_width&quot;:&quot;boxed&quot;}\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-6b0de3d elementor-widget elementor-widget-heading\" data-id=\"6b0de3d\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<h3 class=\"elementor-heading-title elementor-size-default\">L\u2019IA qui cr\u00e9e de la IA<\/h3>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-76552df elementor-widget elementor-widget-text-editor\" data-id=\"76552df\" 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<p><span style=\"font-weight: 400;\">La recherche automatique d\u2019architecture neuronale a quatre axes principaux ou blocs de construction dont elle se sert lors de la cr\u00e9ation d\u2019un r\u00e9seau de neurones :<\/span><\/p><ol><li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\"><strong>L\u2019espace de recherche<\/strong> : il s\u2019agit de l\u2019ensemble des architectures candidates parmi lesquelles on peut choisir, ainsi que les op\u00e9rateurs qui les composent.<\/span><\/li><li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\"><strong>La strat\u00e9gie de recherche<\/strong> : elle guide la fa\u00e7on dont la recherche s\u2019effectue. Elle fait un \u00e9chantillonnage parmi les architectures propos\u00e9es sans construire ou entra\u00eener le mod\u00e8le de sorte qu\u2019on puisse avoir une estimation de la performance th\u00e9orique du mod\u00e8le.<\/span><\/li><li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\"><strong>L&rsquo;algorithme de recherche<\/strong> : il re\u00e7oit les performances estim\u00e9es et optimise les param\u00e8tres pour avoir des candidats qui sont plus performants.<\/span><\/li><li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\"><strong>L&rsquo;\u00e9valuation<\/strong> : les mod\u00e8les sont cr\u00e9\u00e9s par la NAS et ils sont \u00e9valu\u00e9s et compar\u00e9s.<\/span><\/li><\/ol>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-e006906 elementor-widget elementor-widget-text-editor\" data-id=\"e006906\" 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<p><span style=\"font-weight: 400;\">Pour mieux comprendre l&rsquo;interaction entre ces axes, on peut observer le diagramme suivant.<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-a81214c elementor-widget elementor-widget-image\" data-id=\"a81214c\" 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\t\t\t<img decoding=\"async\" width=\"1180\" height=\"560\" src=\"https:\/\/datascientest.com\/wp-content\/uploads\/2023\/10\/recherche-automatique-d-architecture-neuronale3.jpg\" class=\"attachment-full size-full wp-image-219619\" alt=\"\" srcset=\"https:\/\/datascientest.com\/wp-content\/uploads\/2023\/10\/recherche-automatique-d-architecture-neuronale3.jpg 1180w, https:\/\/datascientest.com\/wp-content\/uploads\/2023\/10\/recherche-automatique-d-architecture-neuronale3-300x142.jpg 300w, https:\/\/datascientest.com\/wp-content\/uploads\/2023\/10\/recherche-automatique-d-architecture-neuronale3-1024x486.jpg 1024w, https:\/\/datascientest.com\/wp-content\/uploads\/2023\/10\/recherche-automatique-d-architecture-neuronale3-768x364.jpg 768w\" sizes=\"(max-width: 1180px) 100vw, 1180px\" \/>\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<div class=\"elementor-element elementor-element-3907223 elementor-align-center button-principale elementor-widget elementor-widget-button\" data-id=\"3907223\" data-element_type=\"widget\" data-widget_type=\"button.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<div class=\"elementor-button-wrapper\">\n\t\t\t<a class=\"elementor-button elementor-button-link elementor-size-sm elementor-animation-grow\" href=\"https:\/\/datascientest.com\/formation-intelligence-artificielle\">\n\t\t\t\t\t\t<span class=\"elementor-button-content-wrapper\">\n\t\t\t\t\t\t<span class=\"elementor-button-text\">Se former au Machine Learning<\/span>\n\t\t<\/span>\n\t\t\t\t\t<\/a>\n\t\t<\/div>\n\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-4c06288 e-flex e-con-boxed e-con\" data-id=\"4c06288\" data-element_type=\"container\" data-settings=\"{&quot;content_width&quot;:&quot;boxed&quot;}\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-f87c096 elementor-widget elementor-widget-heading\" data-id=\"f87c096\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<h3 class=\"elementor-heading-title elementor-size-default\">Les diff\u00e9rents types d\u2019approches<\/h3>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-fac2a7f elementor-widget elementor-widget-heading\" data-id=\"fac2a7f\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<h5 class=\"elementor-heading-title elementor-size-medium\">Apprentissage par renforcement<\/h5>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-09d789c elementor-widget elementor-widget-text-editor\" data-id=\"09d789c\" 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<p><span style=\"font-weight: 400;\">Les toutes premi\u00e8res approches ont fait appel \u00e0 <a href=\"https:\/\/datascientest.com\/q-learning-le-machine-learning-avec-apprentissage-par-renforcement\">l\u2019apprentissage par renforcement<\/a>. Dans ce type d\u2019approche l\u2019espace de recherche est consid\u00e9r\u00e9 comme un environnement dans lequel on peut \u00e9voluer et o\u00f9 l\u2019on attribue une r\u00e9compense qui est calcul\u00e9e selon la <\/span><b>m\u00e9thode d\u2019\u00e9valuation des performances<\/b><span style=\"font-weight: 400;\"> de l\u2019architecture cr\u00e9\u00e9e. Cependant, cette m\u00e9thode pr\u00e9sente quelques d\u00e9savantages : notamment les ressources n\u00e9cessaires pour obtenir une architecture suffisamment performante, des <a href=\"https:\/\/datascientest.com\/gpu-data-science\">milliers d\u2019heures-GPU<\/a> sont n\u00e9cessaires pour ce faire.<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-7571b25 elementor-widget elementor-widget-heading\" data-id=\"7571b25\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<h4 class=\"elementor-heading-title elementor-size-medium\">Descente de gradient<\/h4>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-624674e elementor-widget elementor-widget-text-editor\" data-id=\"624674e\" 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<p><span style=\"font-weight: 400;\">Une autre fa\u00e7on d\u2019effectuer une recherche d\u2019architecture neuronale diff\u00e9rentiable est d\u2019optimiser un certain nombre de param\u00e8tres via une <\/span><b>descente de gradient stochastique<\/b><span style=\"font-weight: 400;\">. Dans ce cas les param\u00e8tres, aussi appel\u00e9s poids (des variables continues d\u00e9not\u00e9s par \ud835\udefc), repr\u00e9sentent les op\u00e9rations effectu\u00e9es par l\u2019<\/span><b>architecture candidate<\/b><span style=\"font-weight: 400;\">. Une fois la recherche achev\u00e9e, le mod\u00e8le final est construit en \u00e9chantillonnant les op\u00e9rations constituant l\u2019architecture prenant en compte les poids \ud835\udefc. Les <\/span><b>m\u00e9thodes d\u2019\u00e9chantillonnage<\/b><span style=\"font-weight: 400;\"> peuvent varier, comme par exemple transformer \ud835\udefc en une <\/span><b>distribution de probabilit\u00e9s<\/b><span style=\"font-weight: 400;\"> \u00e0 l\u2019aide d\u2019une fonction softmax.\u00a0<\/span><\/p><p><span style=\"font-weight: 400;\">Additionnellement, les ressources de calcul et temps peuvent \u00eatre davantage \u00e9conomis\u00e9es\u00a0 \u00e0 travers l\u2019impl\u00e9mentation de <\/span><b>m\u00e9canismes de partage des param\u00e8tres<\/b><span style=\"font-weight: 400;\"> de sorte que l\u2019espace de recherche devient partitionn\u00e9 en plusieurs cellules qui repr\u00e9sentent des composantes \u00e9l\u00e9mentaires et \u00e0 partir desquelles l\u2019on peut <strong>construire une architecture neuronale compl\u00e8te<\/strong>. L\u2019avantage de ce m\u00e9canisme provient du fait qu\u2019il n\u2019est plus n\u00e9cessaire de chercher l\u2019int\u00e9gralit\u00e9 de l\u2019architecture sinon quelques parties seulement. En cons\u00e9quence, les m\u00e9thodes diff\u00e9rentiables prennent seulement <\/span><b>quelques dizaines d\u2019heures-GPU<\/b><span style=\"font-weight: 400;\">.<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-d8d8b5e elementor-widget elementor-widget-heading\" data-id=\"d8d8b5e\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<h4 class=\"elementor-heading-title elementor-size-medium\">Algorithmes g\u00e9n\u00e9tiques<\/h4>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-823e849 elementor-widget elementor-widget-text-editor\" data-id=\"823e849\" 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<p><span style=\"font-weight: 400;\">Dans ce cas, on applique le principe darwiniste des <\/span><b>algorithmes g\u00e9n\u00e9tiques<\/b><span style=\"font-weight: 400;\">. Ici l\u2019on consid\u00e8re les architectures candidates comme l\u2019ensemble d\u2019individus qui peuvent se reproduire entre eux en m\u00e9langeant leurs op\u00e9rations (g\u00e8nes) de sorte que cela produise de <\/span><b>nouvelles architectures<\/b><span style=\"font-weight: 400;\">. Ainsi, une population de d\u00e9part est progressivement affin\u00e9e en laissant se reproduire seulement les meilleurs individus s\u00e9lectionn\u00e9s selon une m\u00e9thode d\u2019\u00e9valuation donn\u00e9e. De fa\u00e7on similaire aux <\/span><b>approches d\u2019apprentissage par renforcement<\/b><span style=\"font-weight: 400;\">, ces m\u00e9thodes sont co\u00fbteuses en heures-GPU.<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-3e78b15 elementor-widget elementor-widget-image\" data-id=\"3e78b15\" 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\t\t\t<img decoding=\"async\" width=\"826\" height=\"477\" src=\"https:\/\/datascientest.com\/wp-content\/uploads\/2023\/10\/recherche-automatique-d-architecture-neuronale4.jpg\" class=\"attachment-full size-full wp-image-219639\" alt=\"\" srcset=\"https:\/\/datascientest.com\/wp-content\/uploads\/2023\/10\/recherche-automatique-d-architecture-neuronale4.jpg 826w, https:\/\/datascientest.com\/wp-content\/uploads\/2023\/10\/recherche-automatique-d-architecture-neuronale4-300x173.jpg 300w, https:\/\/datascientest.com\/wp-content\/uploads\/2023\/10\/recherche-automatique-d-architecture-neuronale4-768x444.jpg 768w\" sizes=\"(max-width: 826px) 100vw, 826px\" \/>\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-4ccea65 e-flex e-con-boxed e-con\" data-id=\"4ccea65\" data-element_type=\"container\" data-settings=\"{&quot;content_width&quot;:&quot;boxed&quot;}\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-df2fa66 elementor-widget elementor-widget-heading\" data-id=\"df2fa66\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<h3 class=\"elementor-heading-title elementor-size-default\">Les applications<\/h3>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-a3f4ece elementor-widget elementor-widget-text-editor\" data-id=\"a3f4ece\" 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<p><span style=\"font-weight: 400;\">Les NAS sont un domaine en expansion, qui \u00e9volue \u00e0 une grande vitesse et qui continue \u00e0 d\u00e9passer des architectures d\u00e9velopp\u00e9es \u00e0 la main. Il s\u2019agit aussi d\u2019un domaine avec \u00e9norm\u00e9ment d\u2019applications, pour mentionner quelques unes :\u00a0<\/span><\/p><ul><li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">La classification des images<\/span><\/li><li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">La d\u00e9tection d\u2019objets<\/span><\/li><li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">Le <a href=\"https:\/\/datascientest.com\/introduction-au-nlp-natural-language-processing\">traitement du langage naturel<\/a><\/span><\/li><li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">Le traitement des <a href=\"https:\/\/datascientest.com\/series-temporelles\">s\u00e9ries temporelles<\/a><\/span><\/li><li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">La g\u00e9n\u00e9ration d\u2019images ainsi que de vid\u00e9os ou de sons.<\/span><\/li><\/ul>\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-594c3fb e-flex e-con-boxed e-con\" data-id=\"594c3fb\" data-element_type=\"container\" data-settings=\"{&quot;content_width&quot;:&quot;boxed&quot;}\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-ca6e757 elementor-widget elementor-widget-heading\" data-id=\"ca6e757\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<h3 class=\"elementor-heading-title elementor-size-default\">Avantages et limitations<\/h3>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-da469a2 elementor-widget elementor-widget-text-editor\" data-id=\"da469a2\" 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<p><span style=\"font-weight: 400;\">Une limitation importante des NAS est le co\u00fbt computationnel qu\u2019elles peuvent repr\u00e9senter lorsque l\u2019on essaye d\u2019aborder des sujets complexes avec un <\/span><b>nombre important de solutions latentes<\/b><span style=\"font-weight: 400;\">. Plus que l\u2019espace de recherche est large, plus qu\u2019il y aura des options \u00e0 tester, entra\u00eener et \u00e9valuer il y en aura.\u00a0<\/span><\/p><p><span style=\"font-weight: 400;\">Il est aussi important de noter qu\u2019il est difficile de pr\u00e9dire la performance d\u2019un mod\u00e8le potentiel lorsqu\u2019on l\u2019\u00e9value dans des donn\u00e9es r\u00e9elles.\u00a0<\/span><\/p><p><span style=\"font-weight: 400;\">De plus, l\u2019espace de recherche doit toujours \u00eatre d\u00e9fini \u00e0 la main. Par contre, la ma\u00eetrise sur le sujet n\u2019est plus un emp\u00eachement pour l\u2019<\/span><b>efficacit\u00e9 de l\u2019architecture<\/b><span style=\"font-weight: 400;\">.<\/span><\/p><p><span style=\"font-weight: 400;\">N\u00e9anmoins, ces difficult\u00e9s sont en voie de dispara\u00eetre avec l\u2019arriv\u00e9e des m\u00e9thodes plus rapides et plus compl\u00e8tes d\u2019\u00e9valuation d\u2019architectures. Cela veut dire que bient\u00f4t\u00a0 la recherche automatique d\u2019<\/span><b>architecture neuronale<\/b><span style=\"font-weight: 400;\"> sera facilement applicable dans l\u2019ensemble des entreprises, trouvant des solutions selon leurs besoins de fa\u00e7on flexible et efficace.<\/span><\/p>\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-bb47a80 e-flex e-con-boxed e-con\" data-id=\"bb47a80\" data-element_type=\"container\" data-settings=\"{&quot;content_width&quot;:&quot;boxed&quot;}\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-6a22a4a elementor-widget elementor-widget-heading\" data-id=\"6a22a4a\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<h3 class=\"elementor-heading-title elementor-size-default\">En conclusion<\/h3>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-6145fad elementor-widget elementor-widget-text-editor\" data-id=\"6145fad\" 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<p><span style=\"font-weight: 400;\">La recherche automatique d\u2019architecture neuronale nous permet de cr\u00e9er des nouveaux mod\u00e8les \u00e0 partir de z\u00e9ro qui sont mieux performants que ceux cr\u00e9\u00e9s \u00e0 la main d\u2019une fa\u00e7on plus rapide et optimale tout en restant flexible aux possibles modifications que l\u2019on pourrait souhaiter de faire \u00e0 la main.<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-2f34b53 elementor-align-center button-principale elementor-widget elementor-widget-button\" data-id=\"2f34b53\" data-element_type=\"widget\" data-widget_type=\"button.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<div class=\"elementor-button-wrapper\">\n\t\t\t<a class=\"elementor-button elementor-button-link elementor-size-sm elementor-animation-grow\" href=\"\/formations-data\">\n\t\t\t\t\t\t<span class=\"elementor-button-content-wrapper\">\n\t\t\t\t\t\t<span class=\"elementor-button-text\">D\u00e9couvrir nos formations<\/span>\n\t\t<\/span>\n\t\t\t\t\t<\/a>\n\t\t<\/div>\n\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-92e1bdc e-flex e-con-boxed e-con\" data-id=\"92e1bdc\" data-element_type=\"container\" data-settings=\"{&quot;content_width&quot;:&quot;boxed&quot;}\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-e97fed6 elementor-widget elementor-widget-html\" data-id=\"e97fed6\" data-element_type=\"widget\" data-widget_type=\"html.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<script type=\"application\/ld+json\">\r\n{\r\n  \"@context\": \"https:\/\/schema.org\",\r\n  \"@type\": \"FAQPage\",\r\n  \"mainEntity\": [{\r\n    \"@type\": \"Question\",\r\n    \"name\": \"D'o\u00f9 provient le LLM ?\",\r\n    \"acceptedAnswer\": {\r\n      \"@type\": \"Answer\",\r\n      \"text\": \"La recherche automatique d\u2019architecture neuronale (NAS, de l\u2019anglais Neural Architecture Search) est n\u00e9e du besoin d\u2019optimiser la cr\u00e9ation et le param\u00e9trage des mod\u00e8les de Machine Learning. La NAS essaye de r\u00e9pondre \u00e0 la question : est-ce possible de cr\u00e9er un algorithme capable de cr\u00e9er des r\u00e9seaux de neurones mieux performants que ceux cr\u00e9\u00e9s et param\u00e9tr\u00e9s \u00e0 la main ?\"\r\n    }\r\n  },{\r\n    \"@type\": \"Question\",\r\n    \"name\": \"Comprendre la recherche automatique d\u2019architecture neuronale\",\r\n    \"acceptedAnswer\": {\r\n      \"@type\": \"Answer\",\r\n      \"text\": \"D\u00e8s sa conception, la recherche automatique d\u2019architecture neuronale constitue un sous-domaine de l\u2019AutoML ou Machine Learning automatis\u00e9; domaine charg\u00e9 d\u2019automatiser les t\u00e2ches fastidieuses et it\u00e9ratives du d\u00e9veloppement de mod\u00e8les de Machine Learning. De sa part, l\u2019optimisation des hyperparam\u00e8tres est un sous-domaine des NAS.\"\r\n    }\r\n  },{\r\n    \"@type\": \"Question\",\r\n    \"name\": \"Apprentissage par renforcement\",\r\n    \"acceptedAnswer\": {\r\n      \"@type\": \"Answer\",\r\n      \"text\": \"Les toutes premi\u00e8res approches ont fait appel \u00e0 l\u2019apprentissage par renforcement. Dans ce type d\u2019approche l\u2019espace de recherche est consid\u00e9r\u00e9 comme un environnement dans lequel on peut \u00e9voluer et o\u00f9 l\u2019on attribue une r\u00e9compense qui est calcul\u00e9e selon la m\u00e9thode d\u2019\u00e9valuation des performances de l\u2019architecture cr\u00e9\u00e9e. Cependant, cette m\u00e9thode pr\u00e9sente quelques d\u00e9savantages : notamment les ressources n\u00e9cessaires pour obtenir une architecture suffisamment performante, des milliers d\u2019heures-GPU sont n\u00e9cessaires pour ce faire.\"\r\n    }\r\n  }]\r\n}\r\n<\/script>\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\t\t\t<\/div>\n\t\t","protected":false},"excerpt":{"rendered":"<p>L\u2019utilisation des mod\u00e8les de Deep Learning devient de plus en plus n\u00e9cessaire pour les entreprises. Cependant, son impl\u00e9mentation repr\u00e9sente un d\u00e9fi en termes de cr\u00e9ation d\u2019un mod\u00e8le qui s\u2019adapte bien aux besoins tout en restant efficace et \u00e9conomique en ressources. D\u2019o\u00f9 le besoin de doter l\u2019IA d&rsquo;une compr\u00e9hension plus profonde et d\u2019une autonomie accrue. Cet [&hellip;]<\/p>\n","protected":false},"author":77,"featured_media":219592,"comment_status":"open","ping_status":"open","sticky":false,"template":"elementor_theme","format":"standard","meta":{"content-type":"","inline_featured_image":false},"categories":[2497],"tags":[],"yoast_head":"<!-- This site is optimized with the Yoast SEO Premium plugin v21.7 (Yoast SEO v21.7) - https:\/\/yoast.com\/wordpress\/plugins\/seo\/ -->\n<title>Recherche automatique d&#039;architecture neuronale : Tout ce qu&#039;il faut savoir<\/title>\n<meta name=\"description\" content=\"D\u00e9couvrez dans cet article les d\u00e9fis, le fonctionnement et les applications potentielles dans divers domaines concernant l\u2019IA qui cr\u00e9e une IA.\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/datascientest.com\/recherche-automatique-darchitecture-neuronale-tout-savoir\" \/>\n<meta property=\"og:locale\" content=\"fr_FR\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Recherche automatique d&#039;architecture neuronale : Tout ce qu&#039;il faut savoir\" \/>\n<meta property=\"og:url\" content=\"https:\/\/datascientest.com\/recherche-automatique-darchitecture-neuronale-tout-savoir\" \/>\n<meta property=\"og:site_name\" content=\"Formation Data Science | DataScientest.com\" \/>\n<meta property=\"article:published_time\" content=\"2024-01-24T06:30:00+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2024-01-19T12:09:54+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/datascientest.com\/wp-content\/uploads\/2023\/10\/recherche-automatique-d-architecture-neuronale.jpg\" \/>\n\t<meta property=\"og:image:width\" content=\"1200\" \/>\n\t<meta property=\"og:image:height\" content=\"600\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/jpeg\" \/>\n<meta name=\"author\" content=\"Raphael Kassel\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"\u00c9crit par\" \/>\n\t<meta name=\"twitter:data1\" content=\"Raphael Kassel\" \/>\n\t<meta name=\"twitter:label2\" content=\"Dur\u00e9e de lecture estim\u00e9e\" \/>\n\t<meta name=\"twitter:data2\" content=\"7 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\/\/datascientest.com\/recherche-automatique-darchitecture-neuronale-tout-savoir\",\"url\":\"https:\/\/datascientest.com\/recherche-automatique-darchitecture-neuronale-tout-savoir\",\"name\":\"Recherche automatique d'architecture neuronale : Tout ce qu'il faut savoir\",\"isPartOf\":{\"@id\":\"https:\/\/datascientest.com\/#website\"},\"datePublished\":\"2024-01-24T06:30:00+00:00\",\"dateModified\":\"2024-01-19T12:09:54+00:00\",\"author\":{\"@id\":\"https:\/\/datascientest.com\/#\/schema\/person\/6eca16f63daaafcbc96e15549b9d4ee1\"},\"description\":\"D\u00e9couvrez dans cet article les d\u00e9fis, le fonctionnement et les applications potentielles dans divers domaines concernant l\u2019IA qui cr\u00e9e une IA.\",\"inLanguage\":\"fr-FR\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/datascientest.com\/recherche-automatique-darchitecture-neuronale-tout-savoir\"]}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/datascientest.com\/#website\",\"url\":\"https:\/\/datascientest.com\/\",\"name\":\"Formation Data Science | DataScientest.com\",\"description\":\"Leader de la formation en Data Science\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/datascientest.com\/?s={search_term_string}\"},\"query-input\":\"required name=search_term_string\"}],\"inLanguage\":\"fr-FR\"},{\"@type\":\"Person\",\"@id\":\"https:\/\/datascientest.com\/#\/schema\/person\/6eca16f63daaafcbc96e15549b9d4ee1\",\"name\":\"Raphael Kassel\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"fr-FR\",\"@id\":\"https:\/\/datascientest.com\/#\/schema\/person\/image\/\",\"url\":\"https:\/\/datascientest.com\/wp-content\/uploads\/2023\/11\/raph.jpeg\",\"contentUrl\":\"https:\/\/datascientest.com\/wp-content\/uploads\/2023\/11\/raph.jpeg\",\"caption\":\"Raphael Kassel\"},\"description\":\"Raphael Kassel est le Head of Data Programs chez DataScientest. Il supervise une \u00e9quipe Data, impl\u00e9mente de nouveaux parcours comme \\\"Chef de projet IA\\\" et \\\"Data Product Manager\\\", et contribue \u00e0 la modernisation des op\u00e9rations du groupe. En collaboration \u00e9troite avec l'\u00e9quipe de d\u00e9veloppement, il adapte l'accompagnement acad\u00e9mique aux \u00e9volutions de l'\u00e9quipe data et \u00e0 l'internationalisation.\",\"sameAs\":[\"https:\/\/www.linkedin.com\/in\/raphaelkassel\/\",\"Data Science\"],\"url\":\"https:\/\/datascientest.com\/author\/raphaelkassel\"}]}<\/script>\n<!-- \/ Yoast SEO Premium plugin. -->","yoast_head_json":{"title":"Recherche automatique d'architecture neuronale : Tout ce qu'il faut savoir","description":"D\u00e9couvrez dans cet article les d\u00e9fis, le fonctionnement et les applications potentielles dans divers domaines concernant l\u2019IA qui cr\u00e9e une IA.","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:\/\/datascientest.com\/recherche-automatique-darchitecture-neuronale-tout-savoir","og_locale":"fr_FR","og_type":"article","og_title":"Recherche automatique d'architecture neuronale : Tout ce qu'il faut savoir","og_url":"https:\/\/datascientest.com\/recherche-automatique-darchitecture-neuronale-tout-savoir","og_site_name":"Formation Data Science | DataScientest.com","article_published_time":"2024-01-24T06:30:00+00:00","article_modified_time":"2024-01-19T12:09:54+00:00","og_image":[{"width":1200,"height":600,"url":"https:\/\/datascientest.com\/wp-content\/uploads\/2023\/10\/recherche-automatique-d-architecture-neuronale.jpg","type":"image\/jpeg"}],"author":"Raphael Kassel","twitter_card":"summary_large_image","twitter_misc":{"\u00c9crit par":"Raphael Kassel","Dur\u00e9e de lecture estim\u00e9e":"7 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/datascientest.com\/recherche-automatique-darchitecture-neuronale-tout-savoir","url":"https:\/\/datascientest.com\/recherche-automatique-darchitecture-neuronale-tout-savoir","name":"Recherche automatique d'architecture neuronale : Tout ce qu'il faut savoir","isPartOf":{"@id":"https:\/\/datascientest.com\/#website"},"datePublished":"2024-01-24T06:30:00+00:00","dateModified":"2024-01-19T12:09:54+00:00","author":{"@id":"https:\/\/datascientest.com\/#\/schema\/person\/6eca16f63daaafcbc96e15549b9d4ee1"},"description":"D\u00e9couvrez dans cet article les d\u00e9fis, le fonctionnement et les applications potentielles dans divers domaines concernant l\u2019IA qui cr\u00e9e une IA.","inLanguage":"fr-FR","potentialAction":[{"@type":"ReadAction","target":["https:\/\/datascientest.com\/recherche-automatique-darchitecture-neuronale-tout-savoir"]}]},{"@type":"WebSite","@id":"https:\/\/datascientest.com\/#website","url":"https:\/\/datascientest.com\/","name":"Formation Data Science | DataScientest.com","description":"Leader de la formation en Data Science","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/datascientest.com\/?s={search_term_string}"},"query-input":"required name=search_term_string"}],"inLanguage":"fr-FR"},{"@type":"Person","@id":"https:\/\/datascientest.com\/#\/schema\/person\/6eca16f63daaafcbc96e15549b9d4ee1","name":"Raphael Kassel","image":{"@type":"ImageObject","inLanguage":"fr-FR","@id":"https:\/\/datascientest.com\/#\/schema\/person\/image\/","url":"https:\/\/datascientest.com\/wp-content\/uploads\/2023\/11\/raph.jpeg","contentUrl":"https:\/\/datascientest.com\/wp-content\/uploads\/2023\/11\/raph.jpeg","caption":"Raphael Kassel"},"description":"Raphael Kassel est le Head of Data Programs chez DataScientest. Il supervise une \u00e9quipe Data, impl\u00e9mente de nouveaux parcours comme \"Chef de projet IA\" et \"Data Product Manager\", et contribue \u00e0 la modernisation des op\u00e9rations du groupe. En collaboration \u00e9troite avec l'\u00e9quipe de d\u00e9veloppement, il adapte l'accompagnement acad\u00e9mique aux \u00e9volutions de l'\u00e9quipe data et \u00e0 l'internationalisation.","sameAs":["https:\/\/www.linkedin.com\/in\/raphaelkassel\/","Data Science"],"url":"https:\/\/datascientest.com\/author\/raphaelkassel"}]}},"_links":{"self":[{"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/posts\/219590"}],"collection":[{"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/users\/77"}],"replies":[{"embeddable":true,"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/comments?post=219590"}],"version-history":[{"count":0,"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/posts\/219590\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/media\/219592"}],"wp:attachment":[{"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/media?parent=219590"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/categories?post=219590"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/tags?post=219590"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}