BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//LORIA - ECPv6.16.2//NONSGML v1.0//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
X-WR-CALNAME:LORIA
X-ORIGINAL-URL:https://www.loria.fr
X-WR-CALDESC:Évènements pour LORIA
REFRESH-INTERVAL;VALUE=DURATION:PT1H
X-Robots-Tag:noindex
X-PUBLISHED-TTL:PT1H
BEGIN:VTIMEZONE
TZID:Europe/Paris
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:20220327T010000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:20221030T010000
END:STANDARD
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:20230326T010000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:20231029T010000
END:STANDARD
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:20240331T010000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:20241027T010000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTART;TZID=Europe/Paris:20230202T133000
DTEND;TZID=Europe/Paris:20230202T150000
DTSTAMP:20260525T203028
CREATED:20230117T105450Z
LAST-MODIFIED:20230123T092943Z
UID:17487-1675344600-1675350000@www.loria.fr
SUMMARY:Colloquium du Loria - David Chavalarias
DESCRIPTION:Next Loria Colloquium will take place on Thursday\, February 2nd at 1:30 pm in the Amphitheater. \nWe will have the pleasure to welcome David Chavalarias for a presentation entitled « L’inévitable affaissement de la démocratie sous l’effet des BigTech » \nAvec la montée en puissance des réseaux sociaux\, une lame de fond s’abat sur les démocraties : le tissu social se déchire\, les opinions sont manipulées\, les élections sont déstabilisées. Si les outils numériques ont représenté une innovation majeure dans la production et la diffusion de savoirs\, ils ont également leurs côtés obscurs : ils donnent les clés de l’influence sociale à tout acteur\, politique ou étatique\, qui souhaiterait asseoir ses idées auprès d’un grand nombre de nos concitoyens. Autour du livre Toxic Data et des résultats du projet Politoscope\, nous décortiquerons les interférences entres mécanismes cognitifs\, nouvelles technologies de machine learning et enjeux géopolitiques qui mènent à une déstabilisation systémiques de nos démocraties puis nous dégagerons des pistes de réflexion et de recherche pour y remédier. \nBio : \nDavid Chavalarias est Directeur de Recherche CNRS au CAMS (EHESS) et Directeur de l’Institut des Systèmes complexes de Paris Île-de-France. Normalien\, docteur de l’École Polytechnique en Sciences Cognitives\, ses recherches portent sur la compréhension de nos comportements collectifs et des dynamiques d’opinion à partir de la modélisation et de l’analyse de données du Web. Concepteur de plusieurs macroscopes\, outils numériques qui sont au social ce que le microsope est au vivant\, il les a mobilisés dans son dernier ouvrage « Toxic Data – comment les réseaux manipulent nos opinions » (Flammarion 2022) pour enquêter sur l’impact des médias numériques sur nos démocraties. Site web : http://chavalarias.org \nLes personnes extérieures au Loria peuvent s’inscrire par email auprès de marie.baron (at) loria.fr avant le 31 janvier.
URL:https://www.loria.fr/event/colloquium-du-loria-david-chavalarias/
LOCATION:Amphithéâtre du Loria
CATEGORIES:Colloquium Loria
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Paris:20230222T093000
DTEND;TZID=Europe/Paris:20230222T113000
DTSTAMP:20260525T203028
CREATED:20230220T105612Z
LAST-MODIFIED:20230220T105612Z
UID:17617-1677058200-1677065400@www.loria.fr
SUMMARY:PhD Defense: Yang You
DESCRIPTION:Yang You (Larsen) will defend his thesis on Wednesday February\, 22nd at 9.30 am in room A008. \nHis presentation will be in English and is entitled « Probabilistic Decision-Making Models for Multi-Agent Systems and Human-Robot Collaboration ». \n\n\nThesis Commitee\n– Reviewers:\nAbdel-Illah MOUADDIB\, Université de Caen Normandie\, France\nAdriana TAPUS\, ENSTA Paris\, France\n– Examiners:\nArmelle Brun\, Université de Lorraine\, France\nCaroline Chanel\, ISAE-SUPAERO\, France\n– Thesis supervisors:\nOlivier Buffet\, INRIA Nancy\, France\nVincent Thomas\, Université de Lorraine\, France\n– Invited:\nRachid Alami\, LAAS-CNRS\, France\n\nAbstract\n\nIn this thesis\, using Markov decision models\, we investigate high-level decision-making (task-level\nplanning) for robotics in two aspects: robot-robot collaboration and human-robot collaboration.\n\nIn robot-robot collaboration (RRC)\, we study the decision problems of multiple robots involved to\nachieve a shared goal collaboratively\, and we use the decentralized partially observable\nMarkov decision process (Dec-POMDP) framework to model such RRC problems. Then\, we propose\ntwo novel algorithms for solving Dec-POMDPs. The first algorithm (Inf-JESP) finds Nash\nequilibrium solutions by iteratively building the best-response policy for each agent until no\nimprovement can be made. To handle infinite-horizon Dec-POMDPs\, we represent each agent’s\npolicy using a finite-state controller. The second algorithm (MC-JESP) extends Inf-JESP with\ngenerative models\, which enables us to scale up to large problems. Through experiments\, we\ndemonstrate our methods are competitive with existing Dec-POMDP solvers.\n\nIn human-robot collaboration (HRC)\, we can only control the robot\, and the robot faces uncertain\nhuman objectives and induced behaviors. Therefore\, we attempt to address the challenge\nof deriving robot policies in HRC\, which are robust to the uncertainties about human behaviors.\nIn this direction\, we discuss possible mental models that can be used to model humans in an HRC\ntask. We propose a general approach to derive\, automatically and without prior knowledge\, a\nmodel of human behaviors based on the assumption that the human could also control the robot.\nFrom here\, we then design two algorithms for computing robust robot policies relying on solving\na robot POMDP\, whose state contains the human’s internal state. The first algorithm operates\noffline and gives a complete robot policy that can be used during the robot’s execution. The\nsecond algorithm is an online method\, i.e.\, it plans the robot’s action at each time step during\nexecution. Compared with the offline approach\, the online method only requires a generative\nmodel and thus can scale up to large problems. Experiments with synthetic and real humans\nare conducted in a simulated environment to evaluate these algorithms. We observe that our\nmethods can provide robust robot decisions despite the uncertainties over human objectives and\nbehaviors.\n\nIn this thesis\, our research for RRC provides a foundation for building best-response policies\nin a partially observable and multi-agent setting\, which serves as an important intermediate step\nfor addressing HRC problems. Moreover\, we provide more flexible algorithms using generative\nmodels in each contribution.
URL:https://www.loria.fr/event/phd-defense-yang-you/
LOCATION:A008
CATEGORIES:Soutenance
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Paris:20230224T100000
DTEND;TZID=Europe/Paris:20230224T120000
DTSTAMP:20260525T203028
CREATED:20230223T092944Z
LAST-MODIFIED:20230223T092944Z
UID:17625-1677232800-1677240000@www.loria.fr
SUMMARY:PhD Defense: Shakeel Ahmad Sheikh
DESCRIPTION:Shakeel Ahmad Sheikh (Multispeech) will defend his thesis on Friday\, February 24th at 10 am in room C005. \nHis presentation is entitled « Deep Learning for Stuttering Detection ». \nJury Members:Reviewer: Corinne Fredouille\, Professor\, University of Avignon\, LIA\, FranceReviewer: Benjamin Lecouteux\, Professor\, University of Grenoble Alpes\, LIG\, FranceExaminer: Armelle Brun\, Professor\, Université de Lorraine\, LORIA\, FranceInvitee: Fabrice Hirsch\, Professor\, University of Paul-Valery Montpellier\, Praxiling\, FranceInvitee: Md Sahidullah\, Ex Research Scientist\, Inria\, FranceDirector of thesis : Slim Ouni\, Associate Professor\, University of Lorraine\, LORIA\, France \n\n\nAbstract:\n \nStuttering is a speech disorder that is most frequently observed among speech impairments and results in the form of core behaviours. The tedious and time-consuming task\nof detecting and analysing speech patterns of persons who stutter (PWS)\, with the goal of rectifying them is often handled manually by speech therapists\, and is biased towards\ntheir subjective beliefs. Moreover\, the ASR systems also fail to recognize the stuttered speech\, which makes it impractical for PWS to access virtual digital assistants such as Siri\, Alexa\, etc.\n \nThis thesis tries to develop audio based  stuttering detection (SD) systems that successfully capture different variabilities from stuttering utterances such as speaking styles\, age\,\naccents\, etc.\, and learns robust stuttering representations with an aim to provide a fair\, consistent\, and unbiased assessment of stuttered speech.\n \nWhile most of the existing SD systems use multiple binary classifiers for each stutter type\, we present a unified multi-class StutterNet capable of detecting multiple stutter types.\nApproaching the class-imbalance problem in stuttering domain\, we investigated the impact of applying weighted loss function\, and\, also presented Multi-contextual (MC) Multi-branch\n(MB) StutterNet to improve the detection performance of minority classes. \n \nExploiting the speaker information with an assumption that the stuttering models should be invariant to meta-data such as speaker information\, we present\, an adversarial\nmulti-task learning (MTL) SD method that learns robust stutter discrimintaive speaker-invariant representations.\n \nDue to paucity of unlabelled data\, the automated SD task is limited in its use of large deep models in capturing different variabilities\, we introduced the first-ever SSL framework\nto SD domain. The SSL framework first trains a feature extractor for a pre-text task using a large quantity of unlabelled non-stuttering audio data to capture these different variabilities\,\nand then applies the learned feature extractor to a downstream SD task using limited labelled stuttering audio data.
URL:https://www.loria.fr/event/phd-defense-shakeel-ahmad-sheikh/
LOCATION:C005
CATEGORIES:Soutenance
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Paris:20230228T090000
DTEND;TZID=Europe/Paris:20230228T180000
DTSTAMP:20260525T203028
CREATED:20230116T150259Z
LAST-MODIFIED:20230116T151013Z
UID:17474-1677574800-1677607200@www.loria.fr
SUMMARY:Journée scientifique autour de l’IA
DESCRIPTION:Mardi 28 février 2023 aura lieu une journée scientifique autour de l’IA dans la Faculté des Sciences et Technologies de l’Université de Lorraine à Vandœuvre-lès-Nancy. \nCette journée est ouverte à tous et doit permettre de rassembler les étudiants\, ingénieurs et enseignants chercheurs de nos universités mais aussi le monde socio-économique de plus en plus intéressé par nos recherches menées dans le domaine de l’Intelligence Artificielle. \nLe programme abordera l’IA en santé\, le NLP en passant par les différentes applications de l’IA et ses enjeux locaux attendus par nos partenaires industriels. La participation des doctorants en IA se fera par le biais de sessions posters et de démonstrations. Cette journée intéressera également les étudiants en masters 2 et les jeunes chercheurs. \n\nÉvénement gratuit\nLien vers le programme\nInscription obligatoire avant le 22 février 2023 sur ce lien\nDeadline des soumissions d’abstract de poster/Demo via le site d’inscription\nDéjeuner offert par Grand Enov+\n\n 
URL:https://www.loria.fr/event/journee-scientifique-autour-de-lia/
LOCATION:Faculté des Sciences et Technologies\, Campus Aiguillettes\, Vandœuvre-lès-Nancy\, 54506\, France
CATEGORIES:Séminaire
END:VEVENT
END:VCALENDAR