Now accepted at ICML 2026! Congrats to Magamed :)

While it is intuitively clear that straighter trajectories should reduce discretization error when integrating an ODE (for instance, in flow matching), I could not find a precise bound. I therefore rewrote the proof of Cauchy-Lipschitz to make this explicit. github.com/gpeyre/Discr...

PS5 beats S&P500. Rethink your investment strategy.

We hosted in Nice @aymericdieuleveut.bsky.social & A. Taylor for the SMAI-MODE 2026 mini-course on Performance Estimation Problems with 9 hours of lectures + practicals in two days All material freely available 👇 performanceestimation.github.io/Tutorial-SMA... Thanks to both of them!

🔎 Journée internationale des droits des femmes : portrait de Simona Rota Nodari, chercheuse en mathématiques au LJAD, à découvrir dans « Ces femmes qui font bouger les lignes » 👉 univ-cotedazur.shorthandstories.com/ces-femmes-q... #8mars #FemmesEnScience #UniversiteCoted'Azur #VilledeNice #CNRS

🧵 New preprint! Towards Understanding Steering Strength w/ M. Taimeskhanov and D. Garreau Activation steering is a popular way to control LLM behavior at inference. But how much should you steer? We provide the first theoretical analysis of the steering strength α. 📄 arxiv.org/abs/2602.02712 1/7

Bullshit is like entropy nowadays it can only increase.

It’s incredible that the growth rate of administrative bullshit outpaces LLM progress, but here we are.

Vu que le chauffage est visiblement un détail, on trouve des alternatives. Startup-nation 🤘

Save the date : on organise un workshop sur les maths de l'IA à Bordeaux les 4-6 novembre. N'hésitez pas ! plus on est de fous plus on rit ! (dans les limites des capacités d'accueil haha) Dates & infos : wmathsia2026.sciencesconf.org

After the 2168 series, I read @fabinou.bsky.social series on building a Quake PC fabiensanglard.net/quake_pc/ (late 90s-style), it brings back so much memories 😍!

I tested Pocket TTS from Kyutai. It's fun to use for generating French sentences: the accent is strong, but it's still perfectly understandable, even though French doesn't seem to be in the training dataset! Beyond that, the model is quite impressive especially knowing it has only 100M parameters

Extension pour soumettre une communication : deadline au 21 janvier 23:59 !

Les inscriptions aux journées MODE 2026 à Nice sont désormais ouvertes. Elles se dérouleront du 18 au 20 mars à l'Hôtel Saint-Paul. Les inscriptions sont ouvertes jusqu'au 1 mars (majoration > 9/02). La deadline pour soumettre une communication est le **15 janvier**.

Mail annuel du CNRS pour savoir si tu as obtenu le prix Nobel ✅

I am at #NeurIPS2025, reach out if you want to chat!

🏅Finalist: Gradient Scarcity in Graph Learning with Bilevel Optimization, by Hashem GHANEM, Samuel Vaiter, Nicolas Keriven 5/n openreview.net/forum?id=10Y...

Happy to be at #NeurIPS in San Diego to present our poster ‘Learning Theory for Kernel Bilevel Optimization’ #3005, Fri at 4:30 p.m. Stop by/ping me to chat, especially about statistics, causality, generative models! Let's connect! Joint w/ E. Pauwels, @samuelvaiter.com, @michael-arbel.bsky.social

I have several offers for Master internships / PhDs on graph ML funded by ERC MALAGA for 2026. Don't hesitate to contact me to apply! All infos here: nkeriven.github.io/malaga/

Drame : je me suis gouré de bus à Massy et là je visite *toute* la startup-nation

Est-ce que certain d'entre vous #ESR sont dans un laboratoire qui gère les missions / conférences avec un système de ticket genre GLPI ou Jira ? Si oui, est-ce que c'est l'apocalypse ou plutôt mieux qu'une gestion traditionnel ? Si non, quel niveau d'enthousiasme ou de dégout cela vous inspire ?

Les journées SMAI-MODE 2026 auront lieu à Nice du 18 au 20 mars 2026, précédées par un mini-cours les 16/17. Vous pouvez déposer votre proposition de contrib sur mode2026.sciencesconf.org Inscription à partir du 1er décembre pour finir le 1 mars (avec un tarif majoré à compter du 1 février).

Now accepted to #NeurIPS2025, congrats to @fareselkhoury.bsky.social for his first ML paper!

Happy to share that I have been promoted to senior research scientist (directeur de recherche (*)) at @cnrs.fr effective today! Grateful to my amazing students, collaborators and mentors who made this journey possible. I will continue my research at the math lab of UniCA @ljad-off.bsky.social.

Je viens de réaliser que mon employeur, sans me donner la possibilité d'acquérir un téléphone pro, exige une politique de sécurité sur mon code de smartphone pour accéder aux emails. Je viens donc de faire ce que j'aurai du faire il y a dix ans : enlever mes mails pros de mon téléphone perso 🤷‍♂️

Now accepted at #NeurIPS2025 :)

#Communiqué 🗞️ La médaille d'or 2025 du CNRS est décernée à Stéphane Mallat, mondialement reconnu pour ses travaux autour des mathématiques appliquées au traitement du signal et à l’intelligence artificielle. 👏 👉 cnrs.fr/fr/presse/en... #TalentsCNRS 🏅

Sur le marché public de ma fac, il y a deux hotels de disponible paris intra-muros (1 mois et demi en avance). Et pas "deux" comme une manière de dire peu, DEUX = D-E-U-X = 2. Putain de marché public à la con. Putain de règles administratives à la con.

20 years later I still love this city!

I'm BEYOND THRILLED to tell you that our first original song is now available on all platforms! Plz share if you like it 🙏🏼🙏🏼😊 Cheers

Congratulations to Dr. Sophie Jaffard @sophiejaffard.bsky.social for a brillant PhD defense! We were very lucky with Patricia to have you as a PhD student, and I am sure that you will have a brillant career ahead.

1/2 Julia Brun (M1) «Modèles particulaires de type Cucker-Smale et limites de champ moyen» gagne le prix K. Johnson 🎉 avec un voyage à Paris et visite de la Maison Poincaré #parité #maths #Nice06 #fillesEtMaths @cnrs-insmi.bsky.social @univcotedazur.bsky.social @cnrscotedazur.bsky.social

A reviewer in my batch submitted a review shorter than a tweet. wtf

AWS chatbot is amazing. Q: I cant login neither I can reset my password because phone number verification is not working. A: If you cant reset your password, login into your account to reset your MFA.

Expedition 33 might be overhyped. Let me add my contribution: This is an incredibile masterpiece.

New paper on the generalization of Flow Matching www.arxiv.org/abs/2506.03719 🤯 Why does flow matching generalize? Did you know that the flow matching target you're trying to learn *can only generate training points*? w @quentinbertrand.bsky.social @annegnx.bsky.social @remiemonet.bsky.social 👇👇👇

I am sure that the number of faculty follows the same curve /s

Ma matinée

It seems other authors* have just reached similar conclusions (large minibatches help for OT-CFM), using entropic-reg OT in place of our bilevel-sliced OT * Michal Klein, Alireza Mousavi-Hosseini, Stephen Zhang, @marcocuturi.bsky.social arxiv.org/abs/2506.05526

I wrote in my notes few years ago a *suboptimal* (and maybe wrong 😅) lower bounds of GD for smooth strongly convex function. Nesterov's proof uses seq w/ values in ℝ^ℕ (so infinite dimensional), whereas I do a similar strategy as in the convex case, ie in ℝ^n 1/3 samuelvaiter.com/intro-to-low...

I play Expedition 33 in story mode, and found it not that easy (*). This evening, I launched the game and was impressed with myself because I was much better at parrying than usual. And then I saw the patch notes ... 🥲 (*) translation: I'm getting slaughtered.

One of my papers with @sophiejaffard.bsky.social just turned one 🎂. No reviews yet, but I’m sure the reviewers are carefully choosing every word (one per day).

Le prochain Data Science Colloquium à l'ENS, jeudi 12 juin, sera donné par David Louapre d'Ubisoft, "What modern AI and neuroscience can bring to non-playing characters in video games". David c'est bien sûr également le vulgarisateur scientifique de www.youtube.com/scienceetonn...

📣 New preprint 📣 **Differentiable Generalized Sliced Wasserstein Plans** w/ L. Chapel @rtavenar.bsky.social We propose a Generalized Sliced Wasserstein method that provides an approximated transport plan and which admits a differentiable approximation. arxiv.org/abs/2505.22049 1/5

They say "emacs bankruptcy". I say Marie Kondo x dotemacs = ❤️

Most (?) misunderstood theorems in pop culture 1. Godel's incompleteness theorem 2. Curry-Howard correspondence 3. Universal approximation of neural networks

I have cleaned a bit my lecture notes on Optimal Transport for Machine Learners arxiv.org/abs/2505.06589

Convolution theorem: Fourier transform of the convolution of two functions (under suitable assumptions) is the product of the Fourier transforms of these two functions. buff.ly/aOamDMF

Dual numbers correspond to the completion of the real line with an nilpotent element ε different from 0. It can be thought as a universal linearization of functions, or as the forward-mode of automatic differentiation. buff.ly/vahZ0Fk

Convergence of iterates does not imply convergence of the derivatives. Nevertheless, Gilbert (1994) proposed an interversion limit-derivative theorem under strong assumption on the spectrum of the derivatives. buff.ly/MhK7qiI