pr\u00e9sentent<\/p>\n\n\n\n
Autoris\u00e9 \u00e0 pr\u00e9senter ses travaux en vue de l\u2019obtention du Doctorat de l’Institut Polytechnique de Paris, pr\u00e9par\u00e9 \u00e0 T\u00e9l\u00e9com SudParis en :<\/p>\n\n\n\n
le MERCREDI 19 NOVEMBRE 2025 \u00e0 14h00<\/p>\n\n\n\n
https:\/\/teams.microsoft.com\/dl\/launcher\/launcher.html?url=%2F_%23%2Fl%2Fmeetup-join%2F19%3Ameeting_NWY3NzIwZWItY2FhYi00YWNjLTkyYTgtYjk1YmEwMWM3YWE4%40thread.v2%2F0%3Fcontext%3D%257b%2522Tid%2522%253a%25222b1be215-5cf1-43de-8bc7-946bdf57f9cd%2522%252c%2522Oid%2522%253a%25223cafa483-f4f1-448e-b2d7-1e675507fe9e%2522%257d%26anon%3Dtrue&type=meetup-join&deeplinkId=cd405daa-06f2-4d5c-97df-e0bb60d46364&directDl=true&msLaunch=true&enableMobilePage=true&suppressPrompt=true<\/a> Membres du jury :<\/strong><\/p>\n\n\n\n M. Noel CRESPI<\/strong>, Full professor, T\u00e9l\u00e9com SudParis, FRANCE – Directeur de these Invit\u00e9 :<\/strong><\/p>\n\n\n\n AJAPAKSHA Praboda<\/strong>, ASSISTANT PROFESSOR, Aberystwyth University<\/p>\n\n\n\n R\u00e9sum\u00e9 :<\/strong><\/p>\n\n\n\n Alors que les syst\u00e8mes de traitement du langage naturel (NLP) s’int\u00e8grent de plus en plus dans les technologies sociales interactives, sensibles aux \u00e9motions et d\u00e9cisionnelles, le besoin d’intelligence \u00e9motionnelle dans les mod\u00e8les de langage devient une priorit\u00e9 essentielle de recherche. Si les techniques traditionnelles d’analyse des sentiments permettent de d\u00e9tecter les \u00e9tats \u00e9motionnels de base, elles ont toutefois une capacit\u00e9 limit\u00e9e \u00e0 interpr\u00e9ter les nuances \u00e9motionnelles, cognitives et psychologiques plus profondes du langage humain. La cr\u00e9ation de syst\u00e8mes \u00e9motionnellement intelligents n\u00e9cessite non seulement une reconnaissance fine des \u00e9motions, mais aussi la capacit\u00e9 de raisonner sur leurs causes sous-jacentes et de personnaliser l’interpr\u00e9tation en fonction des traits de personnalit\u00e9 individuels et des facteurs contextuels. Cette th\u00e8se aborde ces d\u00e9fis en \u00e9tudiant comment les caract\u00e9ristiques \u00e9motionnelles, la mod\u00e9lisation psychologique et les m\u00e9canismes de raisonnement peuvent \u00eatre syst\u00e9matiquement int\u00e9gr\u00e9s dans des mod\u00e8les de langage \u00e0 grande \u00e9chelle afin de permettre une compr\u00e9hension \u00e9motionnelle plus profonde et plus adaptative. Les recherches commencent par explorer le r\u00f4le des \u00e9motions fines dans deux t\u00e2ches linguistiques complexes et socialement importantes : la d\u00e9tection des discours haineux implicites, o\u00f9 la haine est exprim\u00e9e indirectement \u00e0 travers des sous-entendus \u00e9motionnels, et l’identification des valeurs humaines dans les discours argumentatifs, o\u00f9 les valeurs morales et culturelles sont souvent signal\u00e9es par un langage affectif. Dans ces deux t\u00e2ches, l’int\u00e9gration d’indices \u00e9motionnels fins am\u00e9liore les performances et l’interpr\u00e9tabilit\u00e9 du mod\u00e8le. S’appuyant sur ces observations, la th\u00e8se pr\u00e9sente un nouveau cadre qui combine les traits de personnalit\u00e9, d\u00e9riv\u00e9s de mod\u00e8les \u00e9tablis tels que le Big Five et Myers-Briggs, avec une m\u00e9thode de tuning bas\u00e9e sur des prompts afin de personnaliser l’interpr\u00e9tation et le raisonnement \u00e9motionnels. Ce cadre permet aux mod\u00e8les de langages d’ajuster leurs r\u00e9ponses en fonction des profils psychologiques sp\u00e9cifiques \u00e0 chaque utilisateur, ce qui permet une inf\u00e9rence plus pr\u00e9cise des causes \u00e9motionnelles dans des contextes \u00e9motionnellement ambigus ou subjectifs. La th\u00e8se propose \u00e9galement un protocole d’\u00e9valuation adapt\u00e9 \u00e0 l’\u00e9valuation du raisonnement \u00e9motionnel en relation avec la personnalit\u00e9, ainsi qu’\u00e0 des techniques d’apprentissage multit\u00e2che et de contrastive training qui renforcent la repr\u00e9sentation \u00e9motionnelle et contextuelle. Les r\u00e9sultats exp\u00e9rimentaux obtenus \u00e0 partir de plusieurs datasets confirment que les m\u00e9thodes propos\u00e9es conduisent \u00e0 des am\u00e9liorations substantielles en mati\u00e8re de compr\u00e9hension des \u00e9motions, de personnalisation et de capacit\u00e9s de raisonnement. En r\u00e9sum\u00e9, ce travail contribue au d\u00e9veloppement de technologies linguistiques \u00e9motionnellement intelligentes en \u00e9tablissant un lien entre la reconnaissance des \u00e9motions, la mod\u00e9lisation des utilisateurs et le raisonnement, ouvrant la voie \u00e0 des syst\u00e8mes plus sensibles au contexte, socialement responsables et align\u00e9s sur l’exp\u00e9rience \u00e9motionnelle humaine. As natural language processing systems become increasingly embedded in socially interactive, emotionally sensitive, and decision-making technologies, the need for emotional intelligence in language models is becoming a critical research priority. While traditional sentiment analysis techniques can detect basic emotional states, they are limited in their ability to interpret the deeper emotional, cognitive, and psychological nuances of human language. Building emotionally intelligent systems requires not only fine-grained recognition of emotions, but also the ability to reason about their underlying causes and to personalize interpretation according to individual personality traits and contextual factors. This thesis addresses these challenges by investigating how emotional features, psychological modeling, and reasoning mechanisms can be systematically integrated into large-scale language models to enable deeper and more adaptive emotional understanding. The research begins by exploring the role of fine-grained emotions in two complex and socially significant language tasks: detecting implicit hate speech, where hate is expressed indirectly through emotional undertones, and identifying human values in argumentative discourse, where moral and cultural values are often signaled through affective language. In both tasks, incorporating fine-grained emotional cues improves model performance and interpretability. Building on these insights, the thesis introduces a new framework that combines personality traits, derived from established models such as the Big Five and Myers-Briggs, with a prompt-based tuning method to personalize emotional interpretation and reasoning. This framework allows language models to adjust their responses based on user-specific psychological profiles, enabling more accurate emotion cause inference in emotionally ambiguous or subjective contexts. The thesis also contributes to an evaluation protocol tailored to assessing emotion reasoning in relation to personality, along with techniques for multi-task learning and contrastive training that strengthen emotional and contextual representation. Experimental results across multiple datasets confirm that the proposed methods lead to substantial improvements in emotion understanding, personalization, and reasoning capabilities. In summary, this work contributes to the development of emotionally intelligent language technologies by bridging emotion recognition, user modeling, and reasoning, paving the way toward systems that are more context-aware, socially responsible, and aligned with human emotional experience.<\/p>\n","protected":false},"excerpt":{"rendered":" L’Ecole doctorale : Ecole Doctorale de l’Institut Polytechnique de Paris 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 Amir Reza JAFARI TEHRANI Autoris\u00e9 \u00e0 pr\u00e9senter ses travaux en vue de l\u2019obtention du Doctorat de l’Institut Polytechnique de Paris, pr\u00e9par\u00e9 \u00e0 T\u00e9l\u00e9com SudParis […]<\/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":[286,402],"tags":[],"class_list":["post-6964","post","type-post","status-publish","format-standard","hentry","category-fractualites-ennews-fr","category-seminaires-ness-2013-fr","entry"],"_links":{"self":[{"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/posts\/6964","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=6964"}],"version-history":[{"count":1,"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/posts\/6964\/revisions"}],"predecessor-version":[{"id":6965,"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/posts\/6964\/revisions\/6965"}],"wp:attachment":[{"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/media?parent=6964"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/categories?post=6964"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/tags?post=6964"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}
Join the meeting now Meeting ID: 357 646 733 405 45<\/a> Passcode: L7d9im2t<\/p>\n\n\n\n
M. Reza FARAHBAKHSH<\/strong>, Assistant professor, T\u00e9l\u00e9com SudParis, FRANCE – Co-encadrant de these
M. Bruce MAGGS<\/strong>, Full professor, Duke University, ETATS-UNIS – Examinateur
M. \u00c1ngel CUEVAS RUM\u00edN<\/strong>, Associate Professor, Universidad Carlos III de Madrid, ESPAGNE – Examinateur
M. Gareth TYSON<\/strong>, Associate Professor, Hong Kong University of Science and Technology, HONG-KONG – Examinateur
M. Marius PREDA<\/strong>, Associate Professor, T\u00e9l\u00e9com SudParis, FRANCE – Examinateur
Mme Elena CABRIO<\/strong>, Full professor, Universit\u00e9 C\u00f4te d\u2019Azur, FRANCE – Rapporteur
M. Tossapon BOONGOEN<\/strong>, Full professor, Aberystwyth University, ROYAUME-UNI – Rapporteur<\/p>\n\n\n\n\u00ab Vers l’intelligence \u00e9motionnelle dans les t\u00e2ches linguistiques \u00e0 l’aide de grands mod\u00e8les de langage \u00bb<\/h2>\n\n\n\n
pr\u00e9sent\u00e9 par Monsieur Amir Reza JAFARI TEHRANI<\/h2>\n\n\n\n
Abstract :<\/strong><\/p>\n\n\n\n