Lightnews — Scholar-powered news

Samuel Vaiter @samuelvaiter.com · 2h

stune blague ?

Samuel Vaiter @samuelvaiter.com · 7d

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

Samuel Vaiter @samuelvaiter.com · Feb 20

📣 New preprint 📣

Learning Theory for Kernel Bilevel Optimization

w/ @fareselkhoury.bsky.social E. Pauwels @michael-arbel.bsky.social

We provide generalization error bounds for bilevel optimization problems where the inner objective is minimized over a RKHS.

arxiv.org/abs/2502.08457

7

Samuel Vaiter @samuelvaiter.com · 8d

FoNdAmEnTaL

3 3

Samuel Vaiter @samuelvaiter.com · 8d

At the same time, I cannot write this message without telling that I am **very** worried about the current administrative evolution of our profession which is evolving quickly towards a system absolutely not calibrated to our duties with a purely accounting PoV.

1

Samuel Vaiter @samuelvaiter.com · 8d

(*) directeur de recherche is roughly equivalent to full prof w/o mandatory teaching duty in France. This is an incredible tenured civil servant position, and I am deeply grateful for it.

DM me if you have questions about the process to enter CNRS. There are similar position e.g. Inria

1 1

Samuel Vaiter @samuelvaiter.com · 8d

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.

1 17

Samuel Vaiter @samuelvaiter.com · 9d

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 🤷‍♂️

5

Samuel Vaiter @samuelvaiter.com · 9d

Now accepted at #NeurIPS2025 :)

Samuel Vaiter @samuelvaiter.com · Jun 2

📣 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

2 6 18

Reposted by Samuel Vaiter

CNRS @cnrs.fr · 28d

#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 🏅

3 45 98

Samuel Vaiter @samuelvaiter.com · 29d

Les fameux islamo-gauchistes niçois 🤣

1

Samuel Vaiter @samuelvaiter.com · Sep 4

Précisons que l'un des deux a 5,9 sur booking. On remerciera l'agence de voyage pour son choix de qualité.

3

Samuel Vaiter @samuelvaiter.com · Sep 4

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.

1 3

Samuel Vaiter @samuelvaiter.com · Aug 15

20 years later I still love this city!

4

Samuel Vaiter @samuelvaiter.com · Aug 11

NeurIPS reviewing procedure is starting to look more complicated than French tax system.

6

Reposted by Samuel Vaiter

Nicolas Keriven @nicolaskeriven.bsky.social · Aug 5

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

1 1 4

Samuel Vaiter @samuelvaiter.com · Aug 1

BERT-like (arxiv.org/abs/1810.04805) models works by doing "masked" language modeling, so instead of predicting the next token like a causal LLM, you predict an arbitrary set of tokens. In particular, you can train them by masking the first one.

For question gen. : aclanthology.org/D19-5821/

BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding

We introduce a new language representation model called BERT, which stands for Bidirectional Encoder Representations from Transformers. Unlike recent language representation models, BERT is designed t...

arxiv.org

2

Samuel Vaiter @samuelvaiter.com · Aug 1

I still don't understand the point of 10.000 x (4~5) characters to tell "Sure I will modify 98% of the manuscrit as soon as you accept the paper" for at end just putting an additional figure p.54354 of your appendix.

This is not a rebbutal, this is begging.

1 3

Samuel Vaiter @samuelvaiter.com · Jul 25

I don't know if there are some stats about inter-topic disparities in scores, could be interesting.

Just for the anecdote, this is the first time I have so much high scores in my batch.

2

Samuel Vaiter @samuelvaiter.com · Jul 25

My AC batch is a proof of a large deviation principle. I have five (over 13) submissions above 4!
But your recommendation still holds ofc

1 3

Samuel Vaiter @samuelvaiter.com · Jul 21

Imagine if the salumiere had a machine that could slice ham automatically. There’d be nothing left for him to do :(

1

Samuel Vaiter @samuelvaiter.com · Jul 17

congrats tony!

1

Samuel Vaiter @samuelvaiter.com · Jul 9

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 1 14

Reposted by Samuel Vaiter

Mathemarium (LJAD) @mathemarium.bsky.social · Jul 4

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

1 2 8

Samuel Vaiter @samuelvaiter.com · Jul 3

Don't you know that LLMs are merely an application of Pythagorean theorem?

1

Samuel Vaiter @samuelvaiter.com · Jul 3

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

1