{"id":1506,"date":"2020-04-02T11:10:03","date_gmt":"2020-04-02T10:10:03","guid":{"rendered":"https:\/\/datascientest.com\/?p=1506"},"modified":"2023-11-09T14:14:56","modified_gmt":"2023-11-09T14:14:56","slug":"reinforcement-learning-comment-apprendre-a-une-ia-a-jouer-au-snake","status":"publish","type":"post","link":"https:\/\/datascientest.com\/reinforcement-learning-comment-apprendre-a-une-ia-a-jouer-au-snake","title":{"rendered":"Comment apprendre \u00e0 une IA \u00e0 jouer au Snake?"},"content":{"rendered":"\t\t<div data-elementor-type=\"wp-post\" data-elementor-id=\"1506\" class=\"elementor elementor-1506\" data-elementor-post-type=\"post\">\n\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-fd75263 e-flex e-con-boxed e-con\" data-id=\"fd75263\" 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-6ad2915 elementor-widget elementor-widget-heading\" data-id=\"6ad2915\" 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\">Vous avez peut-\u00eatre remarqu\u00e9 que les ordinateurs peuvent d\u00e9sormais apprendre automatiquement \u00e0 jouer aux jeux ATARI.\u00a0Ils battent les champions du monde au jeu de Go, les quadrup\u00e8des simul\u00e9s apprennent \u00e0 courir et \u00e0 sauter, et les robots \u00e0 effectuer des t\u00e2ches de manipulation complexes d\u00e9fiant l'\u00e9tat de l'art. Toutes ses avanc\u00e9es sont dues \u00e0 l'\u00e9ventail des possibilit\u00e9s du Reinforcement Learning.<\/h2>\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-7997ae5 e-flex e-con-boxed e-con\" data-id=\"7997ae5\" 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-b724cf8 elementor-widget elementor-widget-heading\" data-id=\"b724cf8\" 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\">Reinforcement Learning<\/h3>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-3b42b88 elementor-widget elementor-widget-text-editor\" data-id=\"3b42b88\" 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>Le c\u0153ur de l&rsquo;apprentissage par renforcement est un agent, c&rsquo;est-\u00e0-dire un algorithme capable de prendre des d\u00e9cisions, \u00e9voluant dans un environnement.<b> L&rsquo;agent (IA) recherche \u00e0 prendre les d\u00e9cisions qui<\/b> <b>maximisent sa r\u00e9compense future<\/b>.<\/p><p>L&rsquo;apprentissage par renforcement diff\u00e8re de l&rsquo;apprentissage supervis\u00e9. Dans l&rsquo;apprentissage supervis\u00e9, les donn\u00e9es de formation ont le corrig\u00e9 avec lui, de sorte que le mod\u00e8le est entra\u00een\u00e9 avec la bonne r\u00e9ponse elle-m\u00eame tandis que dans l&rsquo;apprentissage par renforcement, il n&rsquo;y a pas de r\u00e9ponse mais l&rsquo;IA d\u00e9cide quoi faire pour effectuer la t\u00e2che donn\u00e9e. En l&rsquo;absence de jeu de donn\u00e9es d&rsquo;entra\u00eenement, l\u2019intelligence artificielle est donc livr\u00e9e \u00e0 elle-m\u00eame. Elle commencera par prendre des d\u00e9cisions totalement al\u00e9atoires et, de r\u00e9compense en r\u00e9compense, d\u00e9veloppera sa propre m\u00e9thode pour accomplir syst\u00e9matiquement la t\u00e2che qui lui est confi\u00e9e.<\/p><p>Ainsi, <b>cette m\u00e9thode d\u2019apprentissage pr\u00e9sente l\u2019avantage fascinant de laisser la machine faire preuve de \u00ab cr\u00e9ativit\u00e9 \u00bb.<\/b><\/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-75c4f90 e-flex e-con-boxed e-con\" data-id=\"75c4f90\" 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-1cec5f9 elementor-widget elementor-widget-heading\" data-id=\"1cec5f9\" 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\">Exemple illustratif<\/h3>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-11ce264 elementor-widget elementor-widget-text-editor\" data-id=\"11ce264\" 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>Pour mieux comprendre ce qu\u2019est l\u2019apprentissage par renforcement, il convient de l\u2019illustrer par un exemple simple. Nous avons un agent (rouge), une r\u00e9compense (jaune) et de nombreux obstacles entre les deux. L&rsquo;agent est cens\u00e9 trouver le <b>meilleur chemin possible <\/b>pour atteindre la r\u00e9compense.<\/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-f8aebf8 elementor-widget elementor-widget-image\" data-id=\"f8aebf8\" 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=\"910\" height=\"408\" src=\"https:\/\/datascientest.com\/wp-content\/uploads\/2020\/04\/snake.png\" class=\"attachment-full size-full wp-image-1532\" alt=\"\" srcset=\"https:\/\/datascientest.com\/wp-content\/uploads\/2020\/04\/snake.png 910w, https:\/\/datascientest.com\/wp-content\/uploads\/2020\/04\/snake-300x135.png 300w, https:\/\/datascientest.com\/wp-content\/uploads\/2020\/04\/snake-768x344.png 768w\" sizes=\"(max-width: 910px) 100vw, 910px\" \/>\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-a2ff664 elementor-widget elementor-widget-text-editor\" data-id=\"a2ff664\" 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>Le but de l&rsquo;agent est d&rsquo;obtenir la r\u00e9compense (la food) et d&rsquo;\u00e9viter de passer sur les obstacles (les trous). Si l&rsquo;agent passe sur un obstacle, il obtient une r\u00e9compense de -1 et le jeu est termin\u00e9. Si l&rsquo;agent passe sur la r\u00e9compense, il obtient une r\u00e9compense de +1 et le jeu est \u00e9galement termin\u00e9.<\/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-2b716af e-flex e-con-boxed e-con\" data-id=\"2b716af\" 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-f385f8e elementor-widget elementor-widget-heading\" data-id=\"f385f8e\" 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\u00e9finitions<\/h3>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-67473e4 elementor-widget elementor-widget-text-editor\" data-id=\"67473e4\" 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>\u00c9tant donn\u00e9 un probl\u00e8me d&rsquo;apprentissage par renforcement avec un environnement et un agent, nous allons utiliser les notations suivantes :<\/p><ul><li><u>Un \u00e9pisode<\/u> : Une partie de jeu. Chaque pas de temps d\u2019un \u00e9pisode est index\u00e9 par <b>t: \ud835\udc61\u2208[0,\ud835\udc47]<\/b> (T repr\u00e9sente le nombre de mouvement que l&rsquo;agent a effectu\u00e9 durant la partie).<\/li><li><u>Un \u00e9tat (state)<\/u> : l&rsquo;ensemble des \u00e9tats possibles s&rsquo;\u00e9crit <b>S = {s\u2208S} <\/b>(position de l&rsquo;agent, position de la r\u00e9compense \u2026)<\/li><li><u>Une action<\/u> : l&rsquo;ensemble des actions possibles <b>A={a\u2208A} <\/b>(gauche, droite, haut, bas)<\/li><li><u>La politique<\/u> : la politique suivie par l&rsquo;agent s&rsquo;\u00e9crit <b>\ud835\udf0b<\/b>.<\/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-339e3a2 e-flex e-con-boxed e-con\" data-id=\"339e3a2\" 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-1d34fe8 elementor-widget elementor-widget-heading\" data-id=\"1d34fe8\" 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\">Q-learning <\/h3>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-560c584 elementor-widget elementor-widget-text-editor\" data-id=\"560c584\" 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>Pour rappel, l\u2019objectif de l\u2019agent est trouver les actions qui maximisent sa r\u00e9compense future. Nous d\u00e9finissons la <b>fonction de valeur Q mesurant la somme (actualis\u00e9) des r\u00e9compenses futures en fonction de son \u00e9tat st et son action at.<\/b>\u00a0<\/p><p>Dans notre exemple, nous choisissons st comme la position de l&rsquo;agent dans le monde (initialement : [5,1]), et les actions sont \u00ab\u00a0droite\u00a0\u00bb, \u00ab\u00a0gauche\u00a0\u00bb, \u00ab\u00a0haut\u00a0\u00bb, \u00ab\u00a0bas\u00a0\u00bb.<\/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-18a3d40 e-flex e-con-boxed e-con\" data-id=\"18a3d40\" 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-dcc7053 elementor-widget elementor-widget-heading\" data-id=\"dcc7053\" 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\">Q-Table<\/h3>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-68e7f0d elementor-widget elementor-widget-text-editor\" data-id=\"68e7f0d\" 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>Un <b>Q-Table est une matrice contenant r\u00e9compense future <\/b>estim\u00e9 Q en fonction de l\u2019ensemble des \u00e9tats et des 4 actions possibles. Lorsque l&rsquo;agent explore l&rsquo;environnement, la table est mise \u00e0 jour. L&rsquo;action avec la plus grande valeur est consid\u00e9r\u00e9e comme la meilleure action \u00e0 effectuer.<\/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-13bf754 elementor-widget elementor-widget-image\" data-id=\"13bf754\" 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=\"252\" height=\"60\" src=\"https:\/\/datascientest.com\/wp-content\/uploads\/2020\/04\/Snake3.png\" class=\"attachment-full size-full wp-image-1534\" alt=\"\" \/>\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-9563b94 elementor-widget elementor-widget-text-editor\" data-id=\"9563b94\" 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>L\u2019impl\u00e9mentation de notre agent s\u2019\u00e9crira alors :<\/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-ce2f272 elementor-widget elementor-widget-shortcode\" data-id=\"ce2f272\" data-element_type=\"widget\" data-widget_type=\"shortcode.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<div class=\"elementor-shortcode\"><script src=\"https:\/\/gist.github.com\/15boehler\/52d11b7576d9f081651be937c9bcf8e5.js\"><\/script><\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-6ad0b14 elementor-widget elementor-widget-text-editor\" data-id=\"6ad0b14\" 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>La nouvelle Q-Table pour l&rsquo;action accomplie est donn\u00e9e par la formule suivante :<\/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-2598fad elementor-widget elementor-widget-image\" data-id=\"2598fad\" 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=\"943\" height=\"116\" src=\"https:\/\/datascientest.com\/wp-content\/uploads\/2020\/04\/Snake4.png\" class=\"attachment-full size-full wp-image-1538\" alt=\"\" srcset=\"https:\/\/datascientest.com\/wp-content\/uploads\/2020\/04\/Snake4.png 943w, https:\/\/datascientest.com\/wp-content\/uploads\/2020\/04\/Snake4-300x37.png 300w, https:\/\/datascientest.com\/wp-content\/uploads\/2020\/04\/Snake4-768x94.png 768w\" sizes=\"(max-width: 943px) 100vw, 943px\" \/>\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-a9c5709 elementor-widget elementor-widget-text-editor\" data-id=\"a9c5709\" 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>L&rsquo;\u00e9tape \ud835\udc61 d&rsquo;un \u00e9pisode va se composer alors selon les m\u00e9canismes suivants :<\/p><ul><li>L&rsquo;agent <b>observe<\/b> l&rsquo;\u00e9tat de l&rsquo;environnement st (ex : position de l\u2019agent).<\/li><li>L&rsquo;agent <b>effectue une action<\/b> at+1 en suivant une politique \ud835\udf0b (ex : aller vers le bas). at+1= \u03c0(st, st-1\u2026 at, at-1)<\/li><li>L&rsquo;environnement r\u00e9agit \u00e0 cette action et <b>g\u00e9n\u00e8re une r\u00e9compense<\/b> rt pour l&rsquo;agent : d\u00e9place l\u2019agent et retourne la r\u00e9compense associ\u00e9e \u00e0 cette action [-1(hole), +1(food), 0(case blanche)].<\/li><li>L&rsquo;entra\u00eenement : <b>actualisation<\/b> de la Q-Table<\/li><\/ul>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-a3fb20b elementor-widget elementor-widget-text-editor\" data-id=\"a3fb20b\" 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>L\u2019entra\u00eenement de notre mod\u00e8le donne :<\/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-cd6fc0e elementor-widget elementor-widget-shortcode\" data-id=\"cd6fc0e\" data-element_type=\"widget\" data-widget_type=\"shortcode.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<div class=\"elementor-shortcode\"><script src=\"https:\/\/gist.github.com\/15boehler\/eb0d48f079a72ca90bd9fde749b666f4.js\"><\/script><\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-d23c7f3 elementor-widget elementor-widget-text-editor\" data-id=\"d23c7f3\" 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>En entra\u00eenant le mod\u00e8le sur 100 \u00e9pisodes (parties) on obtient :<\/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-24e76e6 elementor-widget__width-initial elementor-widget elementor-widget-video\" data-id=\"24e76e6\" data-element_type=\"widget\" data-settings=\"{&quot;video_type&quot;:&quot;hosted&quot;,&quot;autoplay&quot;:&quot;yes&quot;,&quot;controls&quot;:&quot;yes&quot;}\" data-widget_type=\"video.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-video .elementor-widget-container{overflow:hidden;transform:translateZ(0)}.elementor-widget-video .elementor-wrapper{aspect-ratio:var(--video-aspect-ratio)}.elementor-widget-video .elementor-wrapper iframe,.elementor-widget-video .elementor-wrapper video{height:100%;width:100%;display:flex;border:none;background-color:#000}@supports not (aspect-ratio:1\/1){.elementor-widget-video .elementor-wrapper{position:relative;overflow:hidden;height:0;padding-bottom:calc(100% \/ var(--video-aspect-ratio))}.elementor-widget-video .elementor-wrapper iframe,.elementor-widget-video .elementor-wrapper video{position:absolute;top:0;right:0;bottom:0;left:0}}.elementor-widget-video .elementor-open-inline .elementor-custom-embed-image-overlay{position:absolute;top:0;right:0;bottom:0;left:0;background-size:cover;background-position:50%}.elementor-widget-video .elementor-custom-embed-image-overlay{cursor:pointer;text-align:center}.elementor-widget-video .elementor-custom-embed-image-overlay:hover .elementor-custom-embed-play i{opacity:1}.elementor-widget-video .elementor-custom-embed-image-overlay img{display:block;width:100%;aspect-ratio:var(--video-aspect-ratio);-o-object-fit:cover;object-fit:cover;-o-object-position:center center;object-position:center center}@supports not (aspect-ratio:1\/1){.elementor-widget-video .elementor-custom-embed-image-overlay{position:relative;overflow:hidden;height:0;padding-bottom:calc(100% \/ var(--video-aspect-ratio))}.elementor-widget-video .elementor-custom-embed-image-overlay img{position:absolute;top:0;right:0;bottom:0;left:0}}.elementor-widget-video .e-hosted-video .elementor-video{-o-object-fit:cover;object-fit:cover}.e-con-inner>.elementor-widget-video,.e-con>.elementor-widget-video{width:var(--container-widget-width);--flex-grow:var(--container-widget-flex-grow)}<\/style>\t\t<div class=\"e-hosted-video elementor-wrapper elementor-open-inline\">\n\t\t\t\t\t<video class=\"elementor-video\" src=\"https:\/\/datascientest.com\/wp-content\/uploads\/2020\/04\/01_exemple.mp4\" autoplay=\"\" controls=\"\" controlsList=\"nodownload\"><\/video>\n\t\t\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-b9edf02 e-flex e-con-boxed e-con\" data-id=\"b9edf02\" 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-e369a5a elementor-widget elementor-widget-heading\" data-id=\"e369a5a\" 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\">Exemple 2 : ROOM MAP<\/h3>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-ae2509c elementor-widget elementor-widget-text-editor\" data-id=\"ae2509c\" 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>Dans l\u2019exemple introductif, nous \u00e9tions dans un monde o\u00f9 la position initiale de l&rsquo;agent ainsi que celle de la r\u00e9compense \u00e9taient identiques \u00e0 chaque \u00e9pisode. Nous avons \u00e9galement supposer <b>aucune r\u00e9apparition de la r\u00e9compense <\/b>(food).<\/p><p>Dans le monde ROOMS MAP, nous allons faire abstraction de ses hypoth\u00e8ses. L\u2019agent ainsi que la r\u00e9compense\u00a0 auront une position initiale al\u00e9atoire. <b>\u00c0 chaque fois que l\u2019agent atteint une r\u00e9compense, il re\u00e7oit une r\u00e9compense +1 et la food r\u00e9appara\u00eet sur une position al\u00e9atoire.<\/b><\/p><p>Dans le premier exemple, la position de l\u2019agent suffisait \u00e0 distinguer la situation de l\u2019agent dans l&rsquo;environnement. Dans notre nouveau monde, la position de l\u2019agent ne suffit plus \u00e0 distinguer la situation de l\u2019agent puisque la position initiale de la food est choisie al\u00e9atoirement. Une fa\u00e7on de r\u00e9soudre le probl\u00e8me est d\u2019ajouter la position de la food dans nos \u00e9tats.<\/p><p>En choisissant comme \u00e9tats la position de l\u2019agent ainsi que la position de la r\u00e9compense, nous pouvons r\u00e9soudre ce probl\u00e8me de la m\u00eame mani\u00e8re :<\/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-7190731 elementor-widget__width-initial elementor-widget elementor-widget-video\" data-id=\"7190731\" data-element_type=\"widget\" data-settings=\"{&quot;video_type&quot;:&quot;hosted&quot;,&quot;controls&quot;:&quot;yes&quot;}\" data-widget_type=\"video.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<div class=\"e-hosted-video elementor-wrapper elementor-open-inline\">\n\t\t\t\t\t<video class=\"elementor-video\" src=\"https:\/\/datascientest.com\/wp-content\/uploads\/2020\/04\/02_room_map.mp4\" controls=\"\" preload=\"metadata\" controlsList=\"nodownload\"><\/video>\n\t\t\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-d1badc1 e-flex e-con-boxed e-con\" data-id=\"d1badc1\" 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-84aa71f elementor-widget elementor-widget-heading\" data-id=\"84aa71f\" 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\">Le Snake<\/h3>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-64aedb5 elementor-widget elementor-widget-text-editor\" data-id=\"64aedb5\" 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>Dans le monde pr\u00e9c\u00e9dent, l&rsquo;agent ne s&rsquo;agrandit pas quand il mange une r\u00e9compense. Dans cette partie, nous allons faire abstraction de cette hypoth\u00e8se, ce qui revient \u00e0 jouer au jeu Snake.<\/p><p>Prendre toutes les positions du serpent (t\u00eate, parties du corps) comme \u00e9tats peut fonctionner, mais il est n\u00e9cessaire d&rsquo;entra\u00eener le mod\u00e8le tr\u00e8s longtemps en raison du nombre \u00e9lev\u00e9 d&rsquo;\u00e9tats. \u00c9tant donn\u00e9 que cette formation est destin\u00e9e \u00e0 donner une explication rapide et \u00e0 afficher un r\u00e9sultat rapide \u00e0 l&rsquo;utilisateur, cette fa\u00e7on de sauvegarder les \u00e9tats n&rsquo;\u00e9tait pas la meilleure approche.<\/p><p>Pour contourner cette limitation, nous pouvons<b> baser nos \u00e9tats uniquement sur la position relative de la r\u00e9compense (par rapport \u00e0 la t\u00eate du serpent) et la position relative de la section de queue (par rapport \u00e0 la t\u00eate du serpent).<\/b><\/p><p>De cette fa\u00e7on, les \u00e9tats S utilis\u00e9s par l\u2019agent pour pr\u00e9dire l\u2019action suivante seront repr\u00e9sent\u00e9s comme suivant :<\/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-ee5620e elementor-widget elementor-widget-image\" data-id=\"ee5620e\" 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=\"393\" height=\"239\" src=\"https:\/\/datascientest.com\/wp-content\/uploads\/2020\/04\/snake5.png\" class=\"attachment-full size-full wp-image-1535\" alt=\"\" srcset=\"https:\/\/datascientest.com\/wp-content\/uploads\/2020\/04\/snake5.png 393w, https:\/\/datascientest.com\/wp-content\/uploads\/2020\/04\/snake5-300x182.png 300w\" sizes=\"(max-width: 393px) 100vw, 393px\" \/>\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-9888d01 elementor-widget elementor-widget-text-editor\" data-id=\"9888d01\" 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>En choisissant ces \u00e9tats, nous pouvons r\u00e9soudre le probl\u00e8me du Snake de la m\u00eame mani\u00e8re que dans les deux premiers exemples.<\/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-739cf10 elementor-widget__width-initial elementor-widget elementor-widget-video\" data-id=\"739cf10\" data-element_type=\"widget\" data-settings=\"{&quot;video_type&quot;:&quot;hosted&quot;,&quot;controls&quot;:&quot;yes&quot;}\" data-widget_type=\"video.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<div class=\"e-hosted-video elementor-wrapper elementor-open-inline\">\n\t\t\t\t\t<video class=\"elementor-video\" src=\"https:\/\/datascientest.com\/wp-content\/uploads\/2020\/04\/03_snake.mp4\" controls=\"\" preload=\"metadata\" controlsList=\"nodownload\"><\/video>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-d0b0eb7 elementor-widget elementor-widget-text-editor\" data-id=\"d0b0eb7\" 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>En rajoutant des \u00e9tats de proximit\u00e9 (danger \u00e0 gauche\u2026), nous pouvons ajouter plusieurs agents dans le m\u00eame monde :<\/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-65b66ca elementor-widget__width-initial elementor-widget elementor-widget-video\" data-id=\"65b66ca\" data-element_type=\"widget\" data-settings=\"{&quot;video_type&quot;:&quot;hosted&quot;,&quot;controls&quot;:&quot;yes&quot;}\" data-widget_type=\"video.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<div class=\"e-hosted-video elementor-wrapper elementor-open-inline\">\n\t\t\t\t\t<video class=\"elementor-video\" src=\"https:\/\/datascientest.com\/wp-content\/uploads\/2020\/04\/04_double_snake.mp4\" controls=\"\" preload=\"metadata\" controlsList=\"nodownload\"><\/video>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-97f87a2 elementor-widget elementor-widget-text-editor\" data-id=\"97f87a2\" 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>Dans cet article, nous avons entrouvert les possibilit\u00e9s du Reinforcement Learning.\u00a0<\/p><p>L\u2019<strong>algorithme Q Learning<\/strong> donne tr\u00e8s bon r\u00e9sultat quand le nombre d\u2019\u00e9tats n\u2019est pas trop \u00e9lev\u00e9.<\/p><p>Ici, nous avons un peu aiguill\u00e9 notre agent en choisissant une structure d\u2019\u00e9tats qui semble \u00eatre pertinente. Mais est-ce qu\u2019elle l\u2019est vraiment ? Est-ce qu\u2019il ne nous manque pas des \u00e9tats ? Est-ce que nous n&rsquo;avons pas biais\u00e9 notre agent ? C\u2019est pourquoi, la plupart du temps, nous allons pr\u00e9f\u00e9rer donner \u00e0 notre mod\u00e8le une capture d&rsquo;\u00e9cran du jeu comme \u00e9tat.<\/p><p>Pour traiter un nombre d\u2019\u00e9tats tr\u00e8s important, <strong>nous pouvons allier Deep Learning et Q-learning<\/strong> en d\u00e9finissant un r\u00e9seau de neurones retournant une estimation de la r\u00e9compense future Q. Nous appelons cette algorithme le Deep Q-Learning.<\/p><p>Envie de d\u00e9marrer une formation plus d\u00e9taill\u00e9e sur les diff\u00e9rents aspects abord\u00e9s de cet article ? <a href=\"https:\/\/datascientest.com\/formations-data\">Explorez tous nos modules !<\/a>\u00a0<\/p><p>N&rsquo;h\u00e9sitez pas \u00e0 <a href=\"mailto:contact@datascientest.com\">nous contacter pour plus d&rsquo;informations<\/a>,\u00a0ou rejoignez <a href=\"https:\/\/www.linkedin.com\/school\/datascientest\/\">nos r\u00e9seaux sociaux<\/a> pour \u00eatre toujours inform\u00e9 des prochains lancements de bootcamps DataScience !<\/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\t\t\t\t\t<\/div>\n\t\t","protected":false},"excerpt":{"rendered":"<p>Vous avez peut-\u00eatre remarqu\u00e9 que les ordinateurs peuvent d\u00e9sormais apprendre automatiquement \u00e0 jouer aux jeux ATARI.&nbsp;Ils battent les champions du monde au jeu de Go, les quadrup\u00e8des simul\u00e9s apprennent \u00e0 courir et \u00e0 sauter, et les robots \u00e0 effectuer des t\u00e2ches de manipulation complexes d\u00e9fiant l&rsquo;\u00e9tat de l&rsquo;art. Toutes ses avanc\u00e9es sont dues \u00e0 l&rsquo;\u00e9ventail [&hellip;]<\/p>\n","protected":false},"author":77,"featured_media":1556,"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>Reinforcement Learning : Comment apprendre \u00e0 une IA \u00e0 jouer au Snake?<\/title>\n<meta name=\"description\" content=\"Qu&#039;est ce le Reinforcement Learning? Comment celui-ci peut-il nous permettre d&#039;apprendre \u00e0 une IA \u00e0 jouer au fameux jeu Snake ?\" \/>\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\/reinforcement-learning-comment-apprendre-a-une-ia-a-jouer-au-snake\" \/>\n<meta property=\"og:locale\" content=\"fr_FR\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Comment apprendre \u00e0 une IA \u00e0 jouer au Snake?\" \/>\n<meta property=\"og:url\" content=\"https:\/\/datascientest.com\/reinforcement-learning-comment-apprendre-a-une-ia-a-jouer-au-snake\" \/>\n<meta property=\"og:site_name\" content=\"Formation Data Science | DataScientest.com\" \/>\n<meta property=\"article:published_time\" content=\"2020-04-02T10:10:03+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2023-11-09T14:14:56+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/datascientest.com\/wp-content\/uploads\/2020\/04\/IA-GAME.jpg\" \/>\n\t<meta property=\"og:image:width\" content=\"2560\" \/>\n\t<meta property=\"og:image:height\" content=\"1707\" \/>\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\/reinforcement-learning-comment-apprendre-a-une-ia-a-jouer-au-snake\",\"url\":\"https:\/\/datascientest.com\/reinforcement-learning-comment-apprendre-a-une-ia-a-jouer-au-snake\",\"name\":\"Reinforcement Learning : Comment apprendre \u00e0 une IA \u00e0 jouer au Snake?\",\"isPartOf\":{\"@id\":\"https:\/\/datascientest.com\/#website\"},\"datePublished\":\"2020-04-02T10:10:03+00:00\",\"dateModified\":\"2023-11-09T14:14:56+00:00\",\"author\":{\"@id\":\"https:\/\/datascientest.com\/#\/schema\/person\/6eca16f63daaafcbc96e15549b9d4ee1\"},\"description\":\"Qu'est ce le Reinforcement Learning? Comment celui-ci peut-il nous permettre d'apprendre \u00e0 une IA \u00e0 jouer au fameux jeu Snake ?\",\"inLanguage\":\"fr-FR\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/datascientest.com\/reinforcement-learning-comment-apprendre-a-une-ia-a-jouer-au-snake\"]}]},{\"@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":"Reinforcement Learning : Comment apprendre \u00e0 une IA \u00e0 jouer au Snake?","description":"Qu'est ce le Reinforcement Learning? Comment celui-ci peut-il nous permettre d'apprendre \u00e0 une IA \u00e0 jouer au fameux jeu Snake ?","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\/reinforcement-learning-comment-apprendre-a-une-ia-a-jouer-au-snake","og_locale":"fr_FR","og_type":"article","og_title":"Comment apprendre \u00e0 une IA \u00e0 jouer au Snake?","og_url":"https:\/\/datascientest.com\/reinforcement-learning-comment-apprendre-a-une-ia-a-jouer-au-snake","og_site_name":"Formation Data Science | DataScientest.com","article_published_time":"2020-04-02T10:10:03+00:00","article_modified_time":"2023-11-09T14:14:56+00:00","og_image":[{"width":2560,"height":1707,"url":"https:\/\/datascientest.com\/wp-content\/uploads\/2020\/04\/IA-GAME.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\/reinforcement-learning-comment-apprendre-a-une-ia-a-jouer-au-snake","url":"https:\/\/datascientest.com\/reinforcement-learning-comment-apprendre-a-une-ia-a-jouer-au-snake","name":"Reinforcement Learning : Comment apprendre \u00e0 une IA \u00e0 jouer au Snake?","isPartOf":{"@id":"https:\/\/datascientest.com\/#website"},"datePublished":"2020-04-02T10:10:03+00:00","dateModified":"2023-11-09T14:14:56+00:00","author":{"@id":"https:\/\/datascientest.com\/#\/schema\/person\/6eca16f63daaafcbc96e15549b9d4ee1"},"description":"Qu'est ce le Reinforcement Learning? Comment celui-ci peut-il nous permettre d'apprendre \u00e0 une IA \u00e0 jouer au fameux jeu Snake ?","inLanguage":"fr-FR","potentialAction":[{"@type":"ReadAction","target":["https:\/\/datascientest.com\/reinforcement-learning-comment-apprendre-a-une-ia-a-jouer-au-snake"]}]},{"@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\/1506"}],"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=1506"}],"version-history":[{"count":0,"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/posts\/1506\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/media\/1556"}],"wp:attachment":[{"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/media?parent=1506"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/categories?post=1506"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/tags?post=1506"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}