{"id":7681,"date":"2026-09-02T15:38:57","date_gmt":"2026-09-02T13:38:57","guid":{"rendered":"https:\/\/samovar.telecom-sudparis.eu\/?p=7681"},"modified":"2026-09-02T15:38:57","modified_gmt":"2026-09-02T13:38:57","slug":"avis-de-soutenance-de-monsieur-germain-bregeon","status":"publish","type":"post","link":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/2026\/09\/02\/avis-de-soutenance-de-monsieur-germain-bregeon\/","title":{"rendered":"AVIS DE SOUTENANCE de Monsieur Germain BREGEON"},"content":{"rendered":"\n<h2 class=\"wp-block-heading\">L&rsquo;Ecole doctorale : Ecole Doctorale de l&rsquo;Institut Polytechnique de Paris<br><br>et le Laboratoire de recherche SAMOVAR &#8211; Services r\u00e9partis, Architectures, Mod\u00e9lisation, Validation, Administration des R\u00e9seaux<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">pr\u00e9sentent<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">l\u2019AVIS DE SOUTENANCE de Monsieur Germain BREGEON<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">Autoris\u00e9 \u00e0 pr\u00e9senter ses travaux en vue de l\u2019obtention du Doctorat de l&rsquo;Institut Polytechnique de Paris, pr\u00e9par\u00e9 \u00e0 l&rsquo;Institut Polytechnique de Paris T\u00e9l\u00e9com SudParis en :<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">Signal, Images, Automatique et robotique<\/h2>\n\n\n\n<h1 class=\"wp-block-heading\">\u00ab Vers une Biblioth\u00e8que d&rsquo;Apprentissage Profond G\u00e9n\u00e9raliste pour les Maillages Triangulaires : Op\u00e9rateurs, Autoencodeurs et Compression Apprise \u00bb<\/h1>\n\n\n\n<p class=\"wp-block-paragraph\">le MARDI 8 SEPTEMBRE 2026 \u00e0 14h00<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u00e0<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Amphith\u00e9\u00e2tre 11<br>9 Rue Charles Fourier, 91000 \u00c9vry-Courcouronnes<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Membres du jury :<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>M. Titus&nbsp;ZAHARIA<\/strong>, Professeur, Institut Polytechnique de Paris T\u00e9l\u00e9com SudParis, FRANCE &#8211; Directeur de th\u00e8se<br><strong>M. Marius&nbsp;PREDA<\/strong>, Ma\u00eetre de conf\u00e9rences, Institut Polytechnique de Paris T\u00e9l\u00e9com SudParis, FRANCE &#8211; Co-encadrant de th\u00e8se<br><strong>Mme Mokraoui&nbsp;ANISSA<\/strong>, Professeure des universit\u00e9s, Universit\u00e9 Sorbonne Paris Nord &#8211; Paris 13, FRANCE &#8211; Examinateur<br><strong>M. Tian&nbsp;DONG<\/strong>, Docteur, Interdigital, ETATS-UNIS &#8211; Examinateur<br><strong>M. Mohamed&nbsp;DAOUDI<\/strong>, Professeur, IMT Nord Europe, FRANCE &#8211; Rapporteur<br><strong>M. Bertrand&nbsp;DELEZOIDE<\/strong>, Professeur associ\u00e9, Universit\u00e9 Paris Saclay, FRANCE &#8211; Rapporteur<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Invit\u00e9 :<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>M. Radu Ispas<\/strong> &#8211; Encadrant &#8211; Keyrus<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">\u00ab Vers une Biblioth\u00e8que d&rsquo;Apprentissage Profond G\u00e9n\u00e9raliste pour les Maillages Triangulaires : Op\u00e9rateurs, Autoencodeurs et Compression Apprise \u00bb<\/h2>\n\n\n\n<h2 class=\"wp-block-heading\">pr\u00e9sent\u00e9 par Monsieur Germain BREGEON<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>R\u00e9sum\u00e9 :<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Cette th\u00e8se introduit MeshConv3D, un framework d&rsquo;apprentissage profond g\u00e9n\u00e9raliste op\u00e9rant nativement sur des maillages triangulaires irr\u00e9guliers. Fond\u00e9 sur une convolution bas\u00e9e sur les faces, un m\u00e9canisme de pooling parall\u00e8le pr\u00e9servant la topologie et un op\u00e9rateur de d\u00e9pooling \u00e0 m\u00e9moire, MeshConv3D traite des maillages de topologie arbitraire ; surfaces non \u00e9tanches et composantes d\u00e9connect\u00e9es, sans n\u00e9cessit\u00e9 de remaillage ni alignement sur un mod\u00e8le de r\u00e9f\u00e9rence. Valid\u00e9 sur des benchmarks de classification et de segmentation avec des performances proche de l&rsquo;\u00e9tat de l&rsquo;art et une empreinte m\u00e9moire bien inf\u00e9rieure aux m\u00e9thodes concurrentes, le framework est ensuite appliqu\u00e9 \u00e0 la compression de maillages 3D \u00e0 travers deux architectures successives. La premi\u00e8re est un autoencodeur g\u00e9om\u00e9trique produisant des repr\u00e9sentations latentes compactes et structur\u00e9es, surpassant les approches neuronales existantes sur les benchmarks de reconstruction. La seconde \u00e9tend ce design en un codec de maillage appris complet, int\u00e9grant mod\u00e9lisation entropique, g\u00e9n\u00e9ration de flux binaire par Draco (pour l\u2019encodage de la connectivit\u00e9) et un m\u00e9canisme de d\u00e9pooling pr\u00e9dictif pour la reconstruction de la connectivit\u00e9. \u00c0 notre connaissance, il s&rsquo;agit du premier pipeline de compression appris de bout en bout op\u00e9rant nativement sur des maillages triangulaires, \u00e9valu\u00e9 dans le contexte industriel des plateformes de t\u00e9l\u00e9 pr\u00e9sence immersive.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Abstract :<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">This thesis introduces MeshConv3D, a general-purpose deep learning framework operating natively on irregular triangular meshes. Built around a face-based convolution operator, a parallel topology-preserving pooling mechanism, and a memory-based unpooling layer, MeshConv3D handles arbitrary mesh topologies, including non-watertight surfaces and disconnected components, without remeshing or template alignment. Validated on classification and segmentation benchmarks at state-of-the-art accuracy and with significantly lower memory requirements than competing methods, the framework is further applied to 3D mesh compression through two successive architectures. The first is a geometry-focused autoencoder producing compact, semantically structured latent representations that outperform existing neural approaches on standard reconstruction benchmarks. The second extends this design into a complete learned mesh codec, integrating entropy modeling, Draco-based bitstream generation for connectivity, and a novel predictive unpooling mechanism for connectivity reconstruction. To our knowledge, this constitutes the first end-to-end learned compression pipeline operating natively on triangular meshes, evaluated in the industrial context of immersive virtual meeting platforms.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>L&rsquo;Ecole doctorale : Ecole Doctorale de l&rsquo;Institut Polytechnique de Paris et le Laboratoire de recherche SAMOVAR &#8211; Services r\u00e9partis, Architectures, Mod\u00e9lisation, Validation, Administration des R\u00e9seaux pr\u00e9sentent l\u2019AVIS DE SOUTENANCE de Monsieur Germain BREGEON Autoris\u00e9 \u00e0 pr\u00e9senter ses travaux en vue de l\u2019obtention du Doctorat de l&rsquo;Institut Polytechnique de Paris, pr\u00e9par\u00e9 \u00e0 l&rsquo;Institut Polytechnique de Paris [&hellip;]<\/p>\n","protected":false},"author":4,"featured_media":0,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"ocean_post_layout":"","ocean_both_sidebars_style":"","ocean_both_sidebars_content_width":0,"ocean_both_sidebars_sidebars_width":0,"ocean_sidebar":"","ocean_second_sidebar":"","ocean_disable_margins":"enable","ocean_add_body_class":"","ocean_shortcode_before_top_bar":"","ocean_shortcode_after_top_bar":"","ocean_shortcode_before_header":"","ocean_shortcode_after_header":"","ocean_has_shortcode":"","ocean_shortcode_after_title":"","ocean_shortcode_before_footer_widgets":"","ocean_shortcode_after_footer_widgets":"","ocean_shortcode_before_footer_bottom":"","ocean_shortcode_after_footer_bottom":"","ocean_display_top_bar":"default","ocean_display_header":"default","ocean_header_style":"","ocean_center_header_left_menu":"","ocean_custom_header_template":"","ocean_custom_logo":0,"ocean_custom_retina_logo":0,"ocean_custom_logo_max_width":0,"ocean_custom_logo_tablet_max_width":0,"ocean_custom_logo_mobile_max_width":0,"ocean_custom_logo_max_height":0,"ocean_custom_logo_tablet_max_height":0,"ocean_custom_logo_mobile_max_height":0,"ocean_header_custom_menu":"","ocean_menu_typo_font_family":"","ocean_menu_typo_font_subset":"","ocean_menu_typo_font_size":0,"ocean_menu_typo_font_size_tablet":0,"ocean_menu_typo_font_size_mobile":0,"ocean_menu_typo_font_size_unit":"px","ocean_menu_typo_font_weight":"","ocean_menu_typo_font_weight_tablet":"","ocean_menu_typo_font_weight_mobile":"","ocean_menu_typo_transform":"","ocean_menu_typo_transform_tablet":"","ocean_menu_typo_transform_mobile":"","ocean_menu_typo_line_height":0,"ocean_menu_typo_line_height_tablet":0,"ocean_menu_typo_line_height_mobile":0,"ocean_menu_typo_line_height_unit":"","ocean_menu_typo_spacing":0,"ocean_menu_typo_spacing_tablet":0,"ocean_menu_typo_spacing_mobile":0,"ocean_menu_typo_spacing_unit":"","ocean_menu_link_color":"","ocean_menu_link_color_hover":"","ocean_menu_link_color_active":"","ocean_menu_link_background":"","ocean_menu_link_hover_background":"","ocean_menu_link_active_background":"","ocean_menu_social_links_bg":"","ocean_menu_social_hover_links_bg":"","ocean_menu_social_links_color":"","ocean_menu_social_hover_links_color":"","ocean_disable_title":"default","ocean_disable_heading":"default","ocean_post_title":"","ocean_post_subheading":"","ocean_post_title_style":"","ocean_post_title_background_color":"","ocean_post_title_background":0,"ocean_post_title_bg_image_position":"","ocean_post_title_bg_image_attachment":"","ocean_post_title_bg_image_repeat":"","ocean_post_title_bg_image_size":"","ocean_post_title_height":0,"ocean_post_title_bg_overlay":0.5,"ocean_post_title_bg_overlay_color":"","ocean_disable_breadcrumbs":"default","ocean_breadcrumbs_color":"","ocean_breadcrumbs_separator_color":"","ocean_breadcrumbs_links_color":"","ocean_breadcrumbs_links_hover_color":"","ocean_display_footer_widgets":"default","ocean_display_footer_bottom":"default","ocean_custom_footer_template":"","ocean_post_oembed":"","ocean_post_self_hosted_media":"","ocean_post_video_embed":"","ocean_link_format":"","ocean_link_format_target":"self","ocean_quote_format":"","ocean_quote_format_link":"post","ocean_gallery_link_images":"on","ocean_gallery_id":[],"footnotes":""},"categories":[286,169],"tags":[],"class_list":["post-7681","post","type-post","status-publish","format-standard","hentry","category-fractualites-ennews-fr","category-seminaires-armedia","entry"],"_links":{"self":[{"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/posts\/7681","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/users\/4"}],"replies":[{"embeddable":true,"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/comments?post=7681"}],"version-history":[{"count":1,"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/posts\/7681\/revisions"}],"predecessor-version":[{"id":7682,"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/posts\/7681\/revisions\/7682"}],"wp:attachment":[{"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/media?parent=7681"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/categories?post=7681"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/tags?post=7681"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}