David Picard
@davidpicard.bsky.social
2.7K followers 430 following 2.2K posts
Professor of Computer Vision/Machine Learning at Imagine/LIGM, École nationale des Ponts et Chaussées @ecoledesponts.bsky.social Music & overall happiness 🌳🪻 Born well below 350ppm 😬 mostly silly personal views 📍Paris 🔗 https://davidpicard.github.io/
Posts Media Videos Starter Packs
Pinned
davidpicard.bsky.social
🚨Updated: "How far can we go with ImageNet for Text-to-Image generation?"

TL;DR: train a text2image model from scratch on ImageNet only and beat SDXL.

Paper, code, data available! Reproducible science FTW!
🧵👇

📜 arxiv.org/abs/2502.21318
💻 github.com/lucasdegeorg...
💽 huggingface.co/arijitghosh/...
davidpicard.bsky.social
Oui enfin le setup a l'air hyper artificiel quand même...
davidpicard.bsky.social
Ah, so this is what you were talking about! Very interesting.

On this topic, I keep saying to my students: "In the experiments, you should put placeholder figures and start writing the analysis. Essentially, if you're not able to predict results before they arrive, maybe we're not doing science!"
davidpicard.bsky.social
monomarchos.agency
Aujourd’hui, Lecornu a été nommé PM. Ou peut-être hier, je ne sais pas. J’ai reçu un télégramme de l’asile : « Lecornu nommé. Gouvernement bientôt. Sentiments distingués."
Reposted by David Picard
phillipisola.bsky.social
Over the past year, my lab has been working on fleshing out theory + applications of the Platonic Representation Hypothesis.

Today I want to share two new works on this topic:

Eliciting higher alignment: arxiv.org/abs/2510.02425
Unpaired learning of unified reps: arxiv.org/abs/2510.08492

1/9
Reposted by David Picard
monomarchos.agency
Aujourd’hui, Lecornu a été nommé PM. Ou peut-être hier, je ne sais pas. J’ai reçu un télégramme de l’asile : « Lecornu nommé. Gouvernement bientôt. Sentiments distingués."
davidpicard.bsky.social
What is the selling point for dots? Cleaner background with still a bit of help? Maybe I should try dots at some point then. Although I fear my terrible handwriting is going to be even more unreadable without the crutch that are the lines.
davidpicard.bsky.social
You like the dots better than the lines or just blank?
Reposted by David Picard
liberation.fr
Démontant un à un les arguments avancés contre la taxe Zucman, l’économiste américain Joseph Stiglitz décrit l’égalité fiscale et la taxation des ultrariches comme des conditions de «la démocratie et de la solidarité sociale».

➡️ bit.ly/3KC9TnI
Reposted by David Picard
cvprconference.bsky.social
Getting ready to submit to #CVPR2026? Take a moment to review the author guidelines, which include important policy details and NEW updates.

Author Guidelines: cvpr.thecvf.com/Conferences/...
Reposted by David Picard
cnrs.fr
CNRS @cnrs.fr · 2d
🎶 Quel est le point commun entre le târ iranien et les luth européens ? En quoi le banjo américain témoigne-t-il de la traite des esclaves ? Au Musée de la musique, à Paris, les instruments racontent leurs évolutions et leurs voyages au fil de l’histoire...
👉 lejournal.cnrs.fr/videos/lhist...
Reposted by David Picard
Reposted by David Picard
inference.vc
When I applied to PhDs, I could derive the E-M algo
after 5 yrs of uni and I felt I stood out. Many applicants didn’t do much ML before.

This year, Miranda, barely out of high school starts her undergraduate with a NeurIPS spotlight, two AI olympiad medals, having bern to EEML 2x plus AI retreat 😨
Reposted by David Picard
hildekuehne.bsky.social
The International Max Planck Research School for Intelligent Systems #IMPRS is currently accepting PhD applications! Join our elite doctoral program to conduct research in machine learning, robotics, HCI, computer vision, and related fields.
imprs.is.mpg.de/application

Deadline: Nov. 16, 2025
LinkedIn
This link will take you to a page that’s not on LinkedIn
lnkd.in
davidpicard.bsky.social
There is also the converse of paper that just take an existing method, use massive engineering to have it beat SOTA, and never credit the original paper.
eugenevinitsky.bsky.social
Sometimes I’ll see a paper that implements an idea I’ve had but implements it poorly for engineering reasons. But then it’s hard to write a follow up paper because it’s not “novel” anymore
Reposted by David Picard
nicolang.bsky.social
Working on representation learning for Earth Observation?
Come join the discussion at the EurIPS workshop "REO: Advances in Representation Learning for Earth Observation"

Call for papers deadline: October 15, AoE
Workshop site: sites.google.com/view/reoeurips

@euripsconf.bsky.social @esa.int
Reposted by David Picard
ducha-aiki.bsky.social
We are starting with @sattlertorsten.bsky.social . You still can book a teleport to Prague :)