2024 Schedule
June 26, 14:00, Amphithéâtre F (Bâtiment A29)
Speaker. Giuseppe Valenzise, L2S (CNRS, Université Paris-Saclay, CentraleSupélec)
Titile. AI-based compression of visual content: trends, challenges and opportunities
Abstract. In the past few years, the emergence of deep representation learning and generative models has led to significant advances in the fields of visual information content compression, including traditional 2D image and videos and more recent immersive formats such as light fields and 3D point clouds. In this talk, I will present some recent approaches for coding 2D and 3D content. In particular, I will detail some recent deep learning-based coding schemes for 3D point clouds, as well as ongoing work on ultra-low bitrate video conferencing using animation models. I will conclude with current open problems and challenges in the field to stimulate discussion.
April 4, 14:00, Amphithéâtre Jean-Paul Dom
Speaker. Elsa Cazelles, IRIT (CNRS, Toulouse INP, UT3, UT Capitole, UT2)
Title. Computationally-efficient initialisation of Gaussian processes: The generalised variogram method
Abstract. We present a computationally-efficient strategy to find the hyperparameters of a Gaussian process (GP) avoiding the computation of the likelihood function. Motivated by the fact that training a GP via ML is equivalent (on average) to minimising the KL-divergence between the true and learnt model, we set to explore different metrics/divergences among GPs that are computationally inexpensive and provide estimates close to those of ML. In particular, we identify the GP hyperparameters by projecting the empirical covariance or (Fourier) power spectrum onto a parametric family, thus proposing and studying various measures of discrepancy operating on the temporal or frequency domains. Our contribution extends the Variogram method developed by the geostatistics literature and, accordingly, it is referred to as the Generalised Variogram method (GVM). In this talk, we will start with a brief introduction to Gaussian processes, then present the proposed GVM and finally provide experimental validation using synthetic and real-world data.
February 1, 14:00, Amphithéâtre Jean-Paul Dom
Speaker. Jamal Najim, LIGM (CNRS, Université Gustave Eiffel)
Title. Retour d’expérience sur un cours autour de la transition écologique donné à des L1 math/info.
Abstract. En octobre 2022, la Ministre de l'Enseignement Supérieur et de la Recherche, Sylvie Retailleau, a annoncé que tous les étudiants suivront des enseignements spécifiques à la transition écologique au plus tard en 2025. Après avoir rappelé le contexte et la pression grandissante sur l'enseignement supérieur pour former les étudiants à la transition écologique, je présenterai le cours « Réchauffement climatique, biodiversité, pollutions » que j'ai enseigné en 2022/23 et 2023/24 (S1) aux étudiants de L1 Math/Info à l’université Gustave Eiffel. Je détaillerai le plan de ce cours généraliste: information et climatosceptiques, climat, énergie, réchauffement climatique, conséquences et adaptation, biodiversité, etc. Je donnerai des exemples d'activités/exercices proposés et préciserai l’évaluation. Je ferai un retour d'expérience sur le déroulement du cours, ébaucherai les perspectives d'amélioration et de déploiement envisageables et évoquerai les difficultés potentielles de mise en oeuvre d'un tel cours à une plus grande échelle.