Séminaires LFI

Le séminaire de l’équipe LFI est organisé au LIP6 (ou en visio-conférence en période de confinement).
Pour accéder au campus Pierre et Marie Curie à Jussieu, cliquer ici.

Année 2021

  • Séminaire du 2 mars 2021, à 17h: “Developmental Machine Learning, Curiosity and Deep Reinforcement Learning” par Pierre-Yves Oudeyer , Directeur de recherche à l’Inria et responsable de l’équipe FLOWERS à l’Inria et l’Ensta ParisTech.
    Lieu du séminaire : en visio-conférence.
    Lien de connexion Zoom: à venir

    Résumé :
    Current approaches to AI and machine learning are still fundamentally limited in comparison with autonomous learning capabilities of children. What is remarkable is not that some children become world champions in certains games or specialties: it is rather their autonomy, flexibility and efficiency at learning many everyday skills under strongly limited resources of time, computation and energy. And they do not need the intervention of an engineer for each new task (e.g. they do not need someone to provide a new task specific reward function).
    I will present a research program (Kaplan and Oudeyer, 2004; Oudeyer et al., 2007; Gottlieb and Oudeyer, 2019) that has focused on computational modeling of child development and learning mechanisms in the last decade. I will discuss several developmental forces that guide exploration in large real world spaces, starting from the perspective of how algorithmic models can help us understand better how they work in humans, and in return how this opens new approaches to autonomous machine learning.
    In particular, I will discuss models of curiosity-driven autonomous learning, enabling machines to sample and explore their own goals and their own learning strategies, self-organizing a learning curriculum without any external reward or supervision. I will introduce the Intrinsically Motivated Goal Exploration Processes (IMGEPs-) algorithmic framework, and present two families of IMGEPs: population-based IMGEPs (Baranes and Oudeyer, 2013; Forestie et al.,2017) with learned goal spaces (Pere et al., 2018), which have allowed sample efficient learning learning of skill repertoires in real robots, and goal-conditioned Deep RL-based IMGEPs, which enable strong generalization properties when they are modular (Colas et al., 2019), in particular when leveraging the compositionality of language to imagine goals in curiosity-driven exploration (Colas et al., 2020).


    Bio: Dr. Pierre-Yves Oudeyer is Research Director (DR1) at Inria and head of the Inria and Ensta-ParisTech FLOWERS team (France). Before, he has been a permanent researcher in Sony Computer Science Laboratory for 8 years (1999-2007). He studied theoretical computer science at Ecole Normale Supérieure in Lyon, and received his Ph.D. degree in artificial intelligence from the University Paris VI, France. He has been studying lifelong autonomous learning, and the self-organization of behavioural, cognitive and cultural structures, at the frontiers of artificial intelligence, machine learning, cognitive sciences and educational technologies. He has been developing models of intrinsically motivated learning, pioneering curiosity-driven learning algorithms working in real world robots, and developed theoretical frameworks to understand better human curiosity and autonomous learning. He also studied mechanisms enabling machines and humans to discover, invent, learn and evolve communication systems. He has published two books, more than 100 papers in international journals and conferences, holds 8 patents, gave several invited keynote lectures in international conferences, and received several prizes for his work in developmental robotics and on the origins of language. In particular, he is laureate of the Inria-National Academy of Science young researcher prize in computer sciences, and of an ERC Starting Grant EXPLORERS. He is also editor of IEEE CIS Newsletter on Cognitive and Developmental Systems where he organizes interdisciplinary dialogs in cognitive science, AI and robotics, as well as associate editor of IEEE Transactions on Cognitive and Developmental Systems and Frontiers in Neurorobotics. He has been chair of IEEE CIS Technical Committee on Cognitive and Developmental Systems.

    Séminaire organisé conjointement avec l’IEEE France Section Life Members Affinity Group et le chapitre Computational Intelligence de l’IEEE France Section (http://ieee-ci.lip6.fr/).

Année 2020

  • Séminaire du 17 décembre 2020, à 10h: “La similarité dans l’analogie et la métaphore” par Charles Tijus, EA 4004 – Cognitions Humaine et Artificielle, FED 4246 – LUTIN.
    Lieu du séminaire : en visio-conférence.

    Résumé : Dans les systèmes cognitifs naturels ou artificiels, les entités (objets, situations, organisations, événements) sont généralement mises en relation « à long terme » par catégorisation extensive (regrouper en différenciant / différencier en regroupant) et intensive (généraliser en spécifiant / spécifier en généralisant) à des fins de représentation (taxonomie) mais aussi de raisonnement et de prise de décision (ontologie). Dans un tel système de représentation, lorsqu’une nouvelle entité cible (inconnue) doit être représentée et raisonnée, elle doit être rangée avec ses homologues respectant la similarité catégorielle extensive et intensive.

    La pensée analogique et la métaphorique reflètent des compétences d’adaptation, de flexibilité cognitive, de créativité et d’innovation permettant des performances en raisonnement, compréhension, raisonnement, prise de décision et résolution de problème. Il s’agit de pensées pour lesquelles deux entités sont mises en relations « à court terme » en dehors de la similarité catégorielle ; ces dernières relevant de domaines différents. Dans ces deux cas, lorsqu’une entité cible X doit être représentée ou raisonnée, une entité X doit être trouvée pour servir de source à des fins de raisonnement analogique ou métaphorique.

    Le point de vue exposé et développé au séminaire est que la mesure de similarité analogique et métaphorique doit être orientée à la fois par

    • par le processus de mise en relation d’une entité X de départ et d’arrivée (cible, topique) avec une entité Y (source, véhicule) de transformation d’état de la connaissance de X : (i) pour la production de l’analogie ou de la métaphore (connaissant X, parmi toutes les entités candidates, comment trouver Y), (ii) sa compréhension (connaissant X et Y, comment comprendre la relation à Y), (iii) l’apprentissage (ne connaissant pas X, comment apprendre à reconnaître et connaitre X à partir de Y), et

    • par le but de la tâche justifiant la mise en œuvre de ce processus : nature des inférences à produire (comprendre, apprendre, résoudre).

    Une conjecture est que la mesure de similarité reposerait dans certain cas sur la catégorie attributive qui pourrait être générée littéralement, circonstanciellement, par Y pour considérer X.

    Séminaire co-organisé avec le Chapitre Français de l’IEEE Computational Intelligence Society

  • Séminaire du 19 novembre 2020, à 9h: “Deep Learning Networks for Medical Image Analysis: its past, future, and issues” by Pau-Choo (Julia) Chung, National Cheng Kung University (NCKU), Taiwan.
    Location : video conference.
    This talk is organised by the French Chapter of the IEEE Computational Intelligence Society thanks to the Distinguished Lecturer Program of the IEEE Computational Intelligence Society
    Abstract : Recent advancement of image understanding with deep learning neural networks has brought great attraction to those in image analysis into the focus of deep learning networks. While researchers on video/image analysis have jumped on the bandwagon of deep learning networks, medical image analyzers would be the coming followers. The characteristics of medical images are extremely different from those of photos and video images. The application of medical image analysis is also much more critical. For achieving the best effectiveness and feasibility of medical image analysis with deep learning approaches, several issues have to be considered. In this talk we will give a brief overview of the development of neural networks for medical image analysis in the past and the future trends with deep learning. Several issues in regard of the data preparation, techniques, and clinic applications will also be discussed.
    Biography: Pau-Choo (Julia) Chung (S’89-M’91-SM’02-F’08) received the Ph.D. degree in electrical engineering from Texas Tech University, USA, in 1991. She then joined the Department of Electrical Engineering, National Cheng Kung University (NCKU), Taiwan, in 1991 and has become a full professor in 1996. She served as the Head of Department of Electrical Engineering (2011-2014), the Director of Institute of Computer and Communication Engineering (2008-2011), the Vice Dean of College of Electrical Engineering and Computer Science (2011), the Director of the Center for Research of E-life Digital Technology (2005-2008), and the Director of Electrical Laboratory (2005-2008), NCKU. She was elected Distinguished Professor of NCKU in 2005 and received the Distinguished Professor Award of Chinese Institute of Electrical Engineering in 2012. She also served as Program Director of Intelligent Computing Division, Ministry of Science and Technology (2012-2014), Taiwan. She was the Director General of the Department of Information and Technology Education, Ministry of Education (2016-2018). She served the Vice President for Members Activities, IEEE CIS (2015-2018).Dr. Chung’s research interests include computational intelligence, medical image analysis, video analysis, and pattern recognition. Dr. Chung participated in many international conferences and society activities. She served as the program committee member in many international conferences. She served as the Publicity Co-Chair of WCCI 2014, SSCI 2013, SSCI 2011, and WCCI 2010. She served as an Associate Editor of IEEE Transactions on Neural Network and Learning Systems(2013-2015) and the Associate Editor of IEEE Transactions on Biomedical Circuits and Systems.Dr. Chung was the Chair of IEEE Computational Intelligence Society (CIS) (2004-2005) in Tainan Chapter, the Chair of the IEEE Life Science Systems and Applications Technical Committee (2008-2009). She was a member in BoG of CAS Society (2007-2009, 2010-2012). She served as an IEEE CAS Society Distinguished Lecturer (2005-2007) and the Chair of CIS Distinguished Lecturer Program (2012-2013). She served on two terms of ADCOM member of IEEE CIS (2009-2011, 2012-2014), the Chair of IEEE CIS Women in CI (2014). She is a Member of Phi Tau Phi honor society and is an IEEE Fellow since 2008.
  • Séminaire du 29 octobre 2020, à 14h“Contributions à l’étude du transfert analogique”
    par Fadi BADRA, Université Paris 13, actuellement en délégation CNRS au LIP6.Lieu : LIP6, salle 105 (1er étage), couloir 25-26, 4 place Jussieu, 75005 Paris
    Résumé : Si deux appartements sont similaires, il est plausible qu’ils aient des prix similaires”. Bien qu’il soit au coeur de la pensée humaine,
    ce type d’inférence a été relativement peu étudié en informatique, et concevoir un modèle computationnel du transfert analogique reste toujours à ce jour un défi. Dans ce séminaire, je reviendrai sur 5 années de travaux sur la modélisation qualitative du transfert analogique, je présenterai les résultats obtenus, et les enseignements qu’on peut en tirer.
    Séminaire co-organisé avec le Chapitre Français de l’IEEE Computational Intelligence Society


Avant 2020

  • Séminaires DAPA (Données et Apprentissage Artificiel) de 2010 à 2019 : cliquer ici