L’Ecole doctorale : Sciences et Technologies de l’Information et de la Communication
\net le Laboratoire de recherche SAMOVAR – Services r\u00e9partis, Architectures, MOd\u00e9lisation, Validation, Administration des R\u00e9seaux<\/p>\n
pr\u00e9sentent<\/p>\n
l\u2019AVIS DE SOUTENANCE de Monsieur Nicolas GRANGER<\/strong><\/p>\n Autoris\u00e9 \u00e0 pr\u00e9senter ses travaux en vue de l\u2019obtention du Doctorat de l’Universit\u00e9 Paris-Saclay, pr\u00e9par\u00e9 \u00e0 T\u00e9l\u00e9com SudParis en Robotique<\/p>\n \u00ab Mod\u00e9lisation par r\u00e9seaux de neurones profonds pour l’apprentissage continu d’objets et de gestes par un robot \u00bb<\/p>\n Quand:<\/strong> Le jeudi 10 janvier 2019 \u00e0 14h00 Membres du jury :<\/strong><\/p>\n R\u00e9sum\u00e9 :<\/strong><\/p>\n Cette th\u00e8se a pour but de contribuer \u00e0 am\u00e9liorer les interfaces Homme-machine. En particulier, nos appareils devraient r\u00e9pliquer notre capacit\u00e9 \u00e0 traiter contin\u00fbment des flux d’information. Cependant, le domaine de l\u2019apprentissage statistique d\u00e9di\u00e9 \u00e0 la reconnaissance de s\u00e9ries temporelles pose de multiples d\u00e9fis. Nos travaux utilisent la reconnaissance de gestes comme exemple applicatif, ces donn\u00e9es offrent un m\u00e9lange complexe de poses corporelles et de mouvements, encod\u00e9es sous des modalit\u00e9s tr\u00e8s vari\u00e9es. La premi\u00e8re partie de notre travail compare deux mod\u00e8les temporels de l\u2019\u00e9tat de l\u2019art pour la reconnaissance continue sur des s\u00e9quences, plus pr\u00e9cis\u00e9ment l\u2019hybride r\u00e9seau de neurones — mod\u00e8le de Markov cach\u00e9 (NN-HMM) et les r\u00e9seaux de neurones r\u00e9currents bidirectionnels (BD-RNN) avec des unit\u00e9s command\u00e9es par des portes. Pour ce faire, nous avons impl\u00e9ment\u00e9 un environnement de test partag\u00e9 qui est plus favorable \u00e0 une \u00e9tude comparative \u00e9quitable. Nous proposons des ajustements sur les fonctions de co\u00fbt utilis\u00e9es pour entra\u00eener les r\u00e9seaux de neurones et sur les expressions du mod\u00e8le hybride afin de g\u00e9rer un large d\u00e9s\u00e9quilibre des classes de notre base d\u2019apprentissage. Bien que les publications r\u00e9centes semblent privil\u00e9gier l\u2019architecture BD-RNN, nous d\u00e9montrons que l\u2019hybride NN-HMM demeure comp\u00e9titif. Cependant, ce dernier est plus d\u00e9pendant de son mod\u00e8le d’entr\u00e9es pour mod\u00e9liser les ph\u00e9nom\u00e8nes temporels \u00e0 court terme. Enfin, nous montrons que les facteurs de variations appris sur les entr\u00e9es par les deux mod\u00e8les sont inter-compatibles. Dans un second temps, nous pr\u00e9sentons une \u00e9tude de l’apprentissage dit \u00aben un coup\u00bb appliqu\u00e9 aux gestes. Ce paradigme d’apprentissage gagne en attention mais demeure peu abord\u00e9 dans le cas de s\u00e9ries temporelles. Nous proposons une architecture construite autour d\u2019un r\u00e9seau de neurones bidirectionnel. Son efficacit\u00e9 est d\u00e9montr\u00e9e par la reconnaissance de gestes isol\u00e9s issus d\u2019un dictionnaire de langage des signes. \u00c0 partir de ce mod\u00e8le de r\u00e9f\u00e9rence, nous proposons de multiples am\u00e9liorations inspir\u00e9es par des travaux dans des domaines connexes, et nous \u00e9tudions les avantages ou inconv\u00e9nients de chacun.<\/p>\n Abstract :<\/strong><\/p>\n This thesis aims to improve the intuitiveness of human–computer interfaces. In particular, machines should try to replicate human’s ability to process streams of information continuously. However, the sub-domain of Machine Learning dedicated to recognition on time series remains barred by numerous challenges. Our studies use gesture recognition as an exemplar application, gestures intermix static body poses and movements in a complex manner using widely different modalities. The first part of our work compares two state-of-the-art temporal models for continuous sequence recognition, namely Hybrid Neural Network–Hidden Markov Models (NN-HMM) and Bidirectional Recurrent Neural Networks (BDRNN) with gated units. To do so, we reimplemented the two within a shared test-bed which is more amenable to a fair comparative work. We propose adjustments to Neural Network training losses and the Hybrid NN-HMM expressions to accommodate for highly imbalanced data classes. Although recent publications tend to prefer BDRNNs, we demonstrate that Hybrid NN-HMM remain competitive. However, the latter rely significantly on their input layers to model short-term patterns. Finally, we show that input representations learned via both approaches are largely inter-compatible. The second part of our work studies one-shot learning, which has received relatively little attention so far, in particular for sequential inputs such as gestures. We propose a model built around a Bidirectional Recurrent Neural Network. Its effectiveness is demonstrated on the recognition of isolated gestures from a sign language lexicon. We propose several improvements over this baseline by drawing inspiration from related works and evaluate their performances, exhibiting different advantages and disadvantages for each.<\/p>\n","protected":false},"excerpt":{"rendered":" L’Ecole doctorale : Sciences et Technologies de l’Information et de la Communication et le Laboratoire de recherche SAMOVAR – Services r\u00e9partis, Architectures, MOd\u00e9lisation, Validation, Administration des R\u00e9seaux pr\u00e9sentent l\u2019AVIS DE SOUTENANCE de Monsieur Nicolas GRANGER Autoris\u00e9 \u00e0 pr\u00e9senter ses travaux en vue de l\u2019obtention du Doctorat de l’Universit\u00e9 Paris-Saclay, pr\u00e9par\u00e9 \u00e0 T\u00e9l\u00e9com SudParis en Robotique […]<\/p>\n","protected":false},"author":1,"featured_media":1136,"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":[296],"tags":[],"class_list":["post-1137","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-theses-2019-fr","entry","has-media"],"_links":{"self":[{"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/posts\/1137","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\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/comments?post=1137"}],"version-history":[{"count":1,"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/posts\/1137\/revisions"}],"predecessor-version":[{"id":1490,"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/posts\/1137\/revisions\/1490"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/media\/1136"}],"wp:attachment":[{"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/media?parent=1137"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/categories?post=1137"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/tags?post=1137"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}
\nO\u00f9:<\/strong> \u00e0 T\u00e9l\u00e9com SudParis, Salle G09 – 9 rue Charles Fourier, 91120 \u00c9VRY<\/p>\n\n\n
\n M. Mounim A. EL YACOUBI, Directeur d’\u00e9tudes , T\u00e9l\u00e9com SudParis, FRANCE <\/td>\n Directeur de these<\/td>\n<\/tr>\n \n M. Gilles GASSO, Professeur, INSA Rouen, FRANCE<\/td>\n Rapporteur<\/td>\n<\/tr>\n \n M. Fabien MOUTARDE, Professeur, Mines ParisTech, FRANCE<\/td>\n Rapporteur<\/td>\n<\/tr>\n \n M. Herv\u00e9 BREDIN, Charg\u00e9 de Recherche, LIMSI, FRANCE<\/td>\n Examinateur<\/td>\n<\/tr>\n \n Mme Alice CAPLIER, Professeur, Grenoble INP, FRANCE<\/td>\n Examinateur<\/td>\n<\/tr>\n \n Mme Laurence LIKFORMAN-SULEM, Ma\u00eetre de Conf\u00e9rences, Telecom ParisTech\/IDS, FRANCE<\/td>\n Examinateur<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n