Seminars

In addition to our weekly reading group, where we discuss recent work within the team and recent papers, MILES regularly hosts researchers for invited seminars. Feel free to contact us if you would like to give a talk!

2026

  • June 19: Anna Mosolova (Inria Paris), Evaluating LLMs’ knowledge using word senses and general knowledge quizzes (slides)
  • May 11 (Three seminars):
    • Simon Raburin (Sorbonne Université)
    • Rodrigo Maulen Soto (LPSM/Sorbonne Université)
    • Daniele Carbone (LPENS, ENS-PSL)
  • May 7: Eliot Tron (SAMM Paris 1 & Toulouse INP)
  • March 26: Yann McLatchie (UCL, UK), Predictively oriented posteriors
  • January 12:
    • Nicolas Berkouk (CNIL), Sociology of AI
    • Edouard Oyallon (ISIR/Sorbonne Université)

2025

  • November 10: Baptiste Hériard-Dubreuil (Institut Langevin, ESPCI), Imaging in complex media: Computational Approaches and Challenges 
  • November 3: Nilo Schwenke (INRIA, LISN, Paris-Saclay), Kernelization of Natural Gradient Methods for Physics Informed Machine-Learning with connections to Galerkin schemes
  • October 6: Luis Nunes Vicente (Lehigh University, USA), Pareto sensitivity, most-changing sub-fronts, and knee solutions
  • June 27: Edwige Cyffers (ISTA, Austria), Empowering Data Owners in Decentralized Learning
  • June 10: Salar Fattahi (University of Michigan, USA), Finding the needle in the haystack: How gradient descent converges to low-dimensional solutions in over-parameterized models
  • March 4: François Pacaud (Mines Paris-PSL), Large-scale nonlinear optimization on the GPU with MadNLP
  • February 18: Guillaume Garrigos (LPSM/Université Paris-Cité), The Stochastic Polyak Stepsize: A fraudulent but interesting algorithm
  • February 11: Jules Samaran (ENS-PSL & Institut Pasteur), Bridging the gap between cellular modalities with Inverse Optimal Transport
  • February 4: Louise Alamichel (INRIA), Bayesian non parametric mixture models and clustering
  • January 21: Pierre Humbert (LPSM/Sorbonne Université), A tutorial on conformal prediction
  • January 7: Pierre Marion (EPFL, Switzerland), Three stories on deep linear networks

2024

  • December 10: Alessandro de Palma (INRIA Paris), From Neural Network Verification to Efficient Robust Training
  • December 3: Nathan Doumèche (Sorbonne Université), PINNS and PIKL: Statistical insights into physics-informed machine learning
  • November 26: Junjie Yang (Telecom Paris), Enhancing Surrogate Regression Methods for Structured Prediction: An Odyssey with Loss Functions
  • October 8: Eva Feillet (CEA LIST, CentraleSupelec), Analysis and recommendation methods for Class-Incremental Learning
  • October 1st: Antoine Gonon (ENS Lyon), Symmetries in Neural Networks
  • May 30: Manon Verbockhaven and Guillaume Charpiat (INRIA Saclay), Neural Architecture Search using expressivity Bottleneck
  • April 25: Pierre Wolinski (Institut de Mathématiques d’Orsay)
  • April 4: Tam Le (LJK/Université Grenoble Alpes)
  • March 14: Michael O’Neill (University of North Carolina Chapel Hill, USA), Complexity of SQP Methods for Deterministically Constrained Stochastic Optimization

2023

  • December 7: Louis Béthune (IRIT, Université de Toulouse), On Lipschitz neural networks
  • September 4: Minh Ha Quang (RIKEN, Japan), An optimal transport and information geometric framework for Gaussian processes
  • June 29: Albert Berahas (University of Michigan, USA), Next generation algorithms for stochastic optimization with constraints
  • April 25: Elisa Riccietti (ENS Lyon), Multilevel optimization methods and their application to the training of physics informed neural networks
  • April 13: Edouard Oyallon (ISIR/Sorbonne Université), Decoupled Greedy Learning of CNNs for Synchronous and Asynchronous Distributed Learning
  • March 16: Adèle Douin (ENS-PSL), Knit-quakes Prediction through Machine Learning

2022

  • December 8: Raphaël Berthier (EPFL, Switzerland), Incremental learning in diagonal linear networks
  • December 7: Nicolas Schreuder (University of Genova, Italy), Fair statistical learning: a study of the Demographic Parity constraint
  • November 24: Batiste Le Bars (INRIA Lille), Refined Convergence and Topology Learning for Decentralized SGD with Heterogeneous Data
  • November 17: Quentin Barthélémy (FoxStream), Making Convolutional Networks Shift-Invariant Again