SEMINARS

MonthWeekDay
October 2019
Monday Tuesday Wednesday Thursday Friday Saturday Sunday
September 30, 2019 October 1, 2019 October 2, 2019 October 3, 2019 October 4, 2019 October 5, 2019 October 6, 2019
October 7, 2019 October 8, 2019 October 9, 2019 October 10, 2019 October 11, 2019 October 12, 2019 October 13, 2019
October 14, 2019 October 15, 2019 October 16, 2019 October 17, 2019 October 18, 2019 October 19, 2019 October 20, 2019
October 21, 2019 October 22, 2019 October 23, 2019 October 24, 2019 October 25, 2019 October 26, 2019 October 27, 2019
October 28, 2019 October 29, 2019 October 30, 2019 October 31, 2019 November 1, 2019 November 2, 2019 November 3, 2019
  • July 1, 2019 @ ENSIMAG - Amphi D -- Soutenance de thèse Michael Mercier : Contribution à la convergence d'infrastructure entre le calcul haute performance (HPC) et le traitement de données a large échelle (Big Data)

    La quantité de données produites, dans le monde scientifique comme dans le monde commercial, est en constante augmentation. Cela implique un changement d'échelle pour les processus de collecte, de traitement, et de stockage des données. Le domaine du Big Data a été inventé pour traiter ces problèmes. Mais ce domaine a été inventé par les géants du web et son intégration sur des machine de calcul intensif pose de nombreux problèmes. En effet, les gestionnaires de ressources utilisé pour soumettre des travaux sur des super calculateurs et les gestionnaires de ressources des frameworks Big Data sont très différents et ne fonctionnent pas ensemble. De plus, les outils d'analyse pour le Big Data ne savent pas tirer partie du matériel très haute performance. Le sujet de ma thèse est de trouver la meilleur approche pour faire interagir ces deux gestionnaires de ressources et de traiter les différents problèmes soulevés par les mouvement de données et leur ordonnancements.

  • September 12, 2019 @ Bâtiment IMAG (406) -- Partially Observable Markov Decision Processes with Finite Memory, by Bruno Ziliotto (CNRS, Paris)

    (joint work with Krishnendu Chatterjee and Raimundo Saona (IST Austria))

    A Partially Observable Markov Decision Process (POMDP) is a discrete-time repeated decision-problem where at each
    period, the stage payoff depends both on the stage action and on the current state of the world. The state evolves
    stochastically from one stage to the other. The decision-maker does not know the state, but receives a stream of
    signals about it. One example is an investor, that does not know exactly the state of the economy, but learns it while
    taking investment decisions. We consider a long interaction, and prove that the decision-maker has approximately
    optimal strategies that have finite memory, and thus can be implemented by a computer.

  • September 19, 2019 @ Bâtiment IMAG (406) -- Some theory on Bayesian neural networks by Julyan Arbel (Mistis, Grenoble)

    In this talk, we first present seminal works at the basis of the theory of Bayesian neural networks. These include Radford Neal result in the 90s regarding the connexion between Gaussian processes and wide neural networks, and the recent developments of this result to deep neural networks. In a second part, we focus on understanding priors in Bayesian neural networks at the unit level. More specifically, we investigate deep Bayesian neural networks with Gaussian weight priors and a class of ReLU-like nonlinearities. We establish that the induced prior distribution on the units before and after activation becomes increasingly heavy-tailed with the depth of the layer.

  • November 14, 2019 @ Bâtiment IMAG (406) -- Paul Duetting

    TBA

  • November 21, 2019 @ -- vivien Quema

    TBA

  • December 5, 2019 @ -- Compression of scientific Data, by Franck Cappello

Comments are closed.