Pierre Wolinski

Courses

Presentation

I am an associate professor (maître de conférences) in computer science at LAMSADE, Paris Dauphine University – PSL, in the MILES project-team. My research focuses on the theory of deep learning, in particular the optimization and regularization of neural networks.

PhD Thesis

I completed my PhD in Machine Learning in March 2020, supervised by Guillaume Charpiat and Yann Ollivier, at the TAU team, LRI.
Here is my CV: [eng].

Thesis: Structural Learning of Neural Networks [PDF file] [link].
Defense carried out on March 6th, 2020.

Interest

• deep learning, neural networks (NNs);
• theory: optimization, Bayesian/Variational inference;
• hyperparameter search, tuning or removal: neural network pruning, learning rate tuning;
• other: neural tangent kernels, random weights, weight initialization...

Works

• Gathering and Exploiting Higher-Order Information when Training Large Structured Models (ECML PKDD 2025), Pierre Wolinski [link] [arXiv] [code]
• Efficient Neural Networks for Tiny Machine Learning: A Comprehensive Review (ACM TIST 2026), Minh Tri Lê, Pierre Wolinski, Julyan Arbel [link] [arXiv]
• Rethinking Gauss-Newton for learning over-parameterized models (NeurIPS 2023), Michael Arbel, Romain Ménégaux, Pierre Wolinski [link]
• Gaussian Pre-Activations in Neural Networks: Myth or Reality? (TMLR 2025), Pierre Wolinski, Julyan Arbel [link] [arXiv] [code]
• An Equivalence between Bayesian Priors and Penalties in Variational Inference (arXiv preprint, 2020 (presented at ISBA 2022)), Pierre Wolinski, Guillaume Charpiat, Yann Ollivier [link]
• Learning with Random Learning Rates (ECML PKDD 2019), Léonard Blier, Pierre Wolinski, Yann Ollivier [link] [arXiv] [code]
• Asymmetrical Scaling Layers for Stable Network Pruning (Draft, 2020), Pierre Wolinski, Guillaume Charpiat, Yann Ollivier [draft]
• Structural Learning of Neural Networks (PhD thesis, Université Paris-Saclay, 2020), Pierre Wolinski
• Consistance des méthodes RKHS dans le cadre de la minimisation d’un risque convexe (Master thesis, 2015), P. Wolinski, supervised by F. d'Alché-Buc, E. Moulines, F. Roueff [link]

Curriculum Vitae

Career

2024 – now: associate professor (maître de conférences) in Computer Science, MILES project-team, LAMSADE, Paris Dauphine University – PSL, France.

2023 – 2024: post-doctoral researcher in the Probability and Statistics team, Paris-Saclay University, France (supervised by Gilles Blanchard and Christophe Giraud).

2021 – 2023: post-doctoral researcher in the Statify team, LJK, Inria Grenoble-Alpes, France (supervised by Julyan Arbel).

2020 – 2021: post-doctoral researcher at the Department of Statistics, University of Oxford, UK (supervised by Judith Rousseau).

Degrees

2020: PhD in Computer Science, Paris-Saclay University.

2016: Graduate degree in Mathematics (minor in Physics) at the École Normale Supérieure (Paris).

2016: Master degree in Mathematics (Probability and Statistics), Paris-Sud University.

Study

2016 – 2020: PhD student in Computer Science at the TAO/Tau team, LRI, Paris-Sud University/Inria Saclay, France (supervised by Guillaume Charpiat and Yann Ollivier).

2011 – 2016: Math student at the École Normale Supérieure, Paris, France.

2008 – 2011: Physics/Chemistry student in "classe préparatoire".

Various links

HAL, GitHub, DBLP, ORCID.