David Picard
@davidpicard.bsky.social
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/
📍Paris 🔗 https://davidpicard.github.io/
Pinned
David Picard
@davidpicard.bsky.social
· Oct 8
🚨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/...
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/...
Reposted by David Picard
You and the text you need to trim in order to meet the word limit
November 10, 2025 at 7:27 AM
You and the text you need to trim in order to meet the word limit
You can't deny the beauty of a forest in autumn 🍂🍁
November 9, 2025 at 9:01 AM
You can't deny the beauty of a forest in autumn 🍂🍁
Reposted by David Picard
I'm still so confused by this whole dustup.
Why should peer-reviewed papers be posted to pre-print servers?
When did everyone become so committed to churning out position papers?
What happened to the public_html/ directory?
Why should peer-reviewed papers be posted to pre-print servers?
When did everyone become so committed to churning out position papers?
What happened to the public_html/ directory?
We are not "banning" reviews; we are just requiring peer review first. Good review articles are important for the field!
You can’t really blame arXiv for the decision to stop publishing computer science stuff (given the flood of slop) but this is also a textbook example of a global public good being gratuitously degraded www.nature.com/articles/d41...
November 8, 2025 at 7:49 PM
I'm still so confused by this whole dustup.
Why should peer-reviewed papers be posted to pre-print servers?
When did everyone become so committed to churning out position papers?
What happened to the public_html/ directory?
Why should peer-reviewed papers be posted to pre-print servers?
When did everyone become so committed to churning out position papers?
What happened to the public_html/ directory?
Reposted by David Picard
One of those internships is on Gromov $\delta$-hyperbolicity for GNNs, and will be cosupervised together with Nicolas, myself and Laetitia Chapel. Take a look and spread the words !
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/
All infos here: nkeriven.github.io/malaga/
MALAGA: Reinventing the Theory of Machine Learning on Large Graphs (ERC StG)
nkeriven.github.io
November 7, 2025 at 1:45 PM
One of those internships is on Gromov $\delta$-hyperbolicity for GNNs, and will be cosupervised together with Nicolas, myself and Laetitia Chapel. Take a look and spread the words !
This Nature retrospective is quite interesting.
To me, the only solution to the credit assignment problem is obvious: stop believing a single person is responsible for every big discovery. It's an artifact of our monkey brain requiring a face for storage, not the reality of how knowledge progresses.
To me, the only solution to the credit assignment problem is obvious: stop believing a single person is responsible for every big discovery. It's an artifact of our monkey brain requiring a face for storage, not the reality of how knowledge progresses.
"stole Rosalind Franklin's work" has become the new orthodoxy. While she was certainly the victim of sexism from Watson, I think her colleague Wilkins was the real villain. Events 1951-53 well covered in Nature in 2023 www.nature.com/articles/d41...
What Rosalind Franklin truly contributed to the discovery of DNA’s structure
Franklin was no victim in how the DNA double helix was solved. An overlooked letter and an unpublished news article, both written in 1953, reveal that she was an equal player.
www.nature.com
November 8, 2025 at 8:22 AM
This Nature retrospective is quite interesting.
To me, the only solution to the credit assignment problem is obvious: stop believing a single person is responsible for every big discovery. It's an artifact of our monkey brain requiring a face for storage, not the reality of how knowledge progresses.
To me, the only solution to the credit assignment problem is obvious: stop believing a single person is responsible for every big discovery. It's an artifact of our monkey brain requiring a face for storage, not the reality of how knowledge progresses.
Reposted by David Picard
Bienvenue au nouveau site de la section 3 CNRS (ancienne section 7 d'informatique), un peu remanié : section3.cnrs.fr
Vous y trouverez notamment les recommandations sur les concours / critères d'évaluation / médailles CNRS / rapports etc.
Vous y trouverez notamment les recommandations sur les concours / critères d'évaluation / médailles CNRS / rapports etc.
Section 03 du CoNRS
Site non officiel de la section 03 du Comité National de la Recherche Scientifique
section3.cnrs.fr
November 7, 2025 at 4:45 PM
Bienvenue au nouveau site de la section 3 CNRS (ancienne section 7 d'informatique), un peu remanié : section3.cnrs.fr
Vous y trouverez notamment les recommandations sur les concours / critères d'évaluation / médailles CNRS / rapports etc.
Vous y trouverez notamment les recommandations sur les concours / critères d'évaluation / médailles CNRS / rapports etc.
Reposted by David Picard
Do you know someone qualified to be an Area Chair for #ICML2026?🤔
(...maybe yourself?)
If so, then nominate them to be an Area Chair! Nomination closes November 17, 2025, so don't delay!🚀🚀🚀
Nomination form: docs.google.com/forms/d/e/1F...
(...maybe yourself?)
If so, then nominate them to be an Area Chair! Nomination closes November 17, 2025, so don't delay!🚀🚀🚀
Nomination form: docs.google.com/forms/d/e/1F...
November 7, 2025 at 4:45 PM
Do you know someone qualified to be an Area Chair for #ICML2026?🤔
(...maybe yourself?)
If so, then nominate them to be an Area Chair! Nomination closes November 17, 2025, so don't delay!🚀🚀🚀
Nomination form: docs.google.com/forms/d/e/1F...
(...maybe yourself?)
If so, then nominate them to be an Area Chair! Nomination closes November 17, 2025, so don't delay!🚀🚀🚀
Nomination form: docs.google.com/forms/d/e/1F...
Reposted by David Picard
Oops. Ooooooooooooops.
I do hope that nobody has been given or denied a job/promotion based on their SpringerNature citation counts in the past 15 years.
arxiv.org/pdf/2511.01675
h/t @nathlarigaldie.bsky.social
I do hope that nobody has been given or denied a job/promotion based on their SpringerNature citation counts in the past 15 years.
arxiv.org/pdf/2511.01675
h/t @nathlarigaldie.bsky.social
November 7, 2025 at 2:02 PM
Oops. Ooooooooooooops.
I do hope that nobody has been given or denied a job/promotion based on their SpringerNature citation counts in the past 15 years.
arxiv.org/pdf/2511.01675
h/t @nathlarigaldie.bsky.social
I do hope that nobody has been given or denied a job/promotion based on their SpringerNature citation counts in the past 15 years.
arxiv.org/pdf/2511.01675
h/t @nathlarigaldie.bsky.social
I hadn't had a good laugh like that in ages. Big thanks to the economists at the federal reserve.
AI could end scarcity, end humanity - or boost trend growth by 0.2 percentage points
November 7, 2025 at 3:09 PM
I hadn't had a good laugh like that in ages. Big thanks to the economists at the federal reserve.
Reposted by David Picard
AI could end scarcity, end humanity - or boost trend growth by 0.2 percentage points
November 7, 2025 at 1:24 PM
AI could end scarcity, end humanity - or boost trend growth by 0.2 percentage points
Reposted by David Picard
OpenReview? More like BrokenReview, amirite?
November 7, 2025 at 11:57 AM
OpenReview? More like BrokenReview, amirite?
Reposted by David Picard
Due to an unexpected system issues prior to the abstract submission deadline, we are extending author registration and abstract submission by 24h until Nov 07, 2025 (Anywhere on Earth). We are working to restore access as quickly as possible.
November 7, 2025 at 11:44 AM
Due to an unexpected system issues prior to the abstract submission deadline, we are extending author registration and abstract submission by 24h until Nov 07, 2025 (Anywhere on Earth). We are working to restore access as quickly as possible.
It's supposed to be a joke, but it was surprisingly easier for me to get an idea of how tall that is.
![Comic. PERSON 1 with white hat: How tall are you? PERSON 2: 5ft 24cm [caption] When switching to metric, make the process easier by doing it in steps.](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:cz73r7iyiqn26upot4jtjdhk/bafkreifptvo6h6kdpf75j3i443ospzumiii7a3pblah2urjrawmyymh3ga@jpeg)
November 7, 2025 at 6:41 AM
It's supposed to be a joke, but it was surprisingly easier for me to get an idea of how tall that is.
Reposted by David Picard
Read the full report here: tinyurl.com/3fvdkph8
November 6, 2025 at 8:12 PM
Read the full report here: tinyurl.com/3fvdkph8
Reposted by David Picard
After Elsevier cut off the tap for the journal Water Resources and Economics in 2024, the research is flowing again, with the launch of the independent new journal, Journal of Water Economics. Edited by the same team, backed by U. Waterloo. #Econsky 😃
openjournals.uwaterloo.ca/index.php/jo...
openjournals.uwaterloo.ca/index.php/jo...
Journal of Water Economics
The Journal of Water Economics publishes theoretical and empirical research on the economics of water resources. Topics include allocation, governance, pricing, regulation, policy instruments, valuati...
openjournals.uwaterloo.ca
November 4, 2025 at 3:45 AM
After Elsevier cut off the tap for the journal Water Resources and Economics in 2024, the research is flowing again, with the launch of the independent new journal, Journal of Water Economics. Edited by the same team, backed by U. Waterloo. #Econsky 😃
openjournals.uwaterloo.ca/index.php/jo...
openjournals.uwaterloo.ca/index.php/jo...
Reposted by David Picard
The JMLR story and operating model should be widely known in academia as a clear success story for full open access. I have friends in the humanities and pure sciences that have no clue this is even possible
November 5, 2025 at 1:16 AM
The JMLR story and operating model should be widely known in academia as a clear success story for full open access. I have friends in the humanities and pure sciences that have no clue this is even possible
Reposted by David Picard
Want to work on generative models and Earth Observation? 🌍
I'm looking for:
🧑💻 an intern on generative models for change detection
🧑🔬 a PhD student on neurosymbolic generative models for geospatial data
Both starting beginning of 2026.
Details are below, feel free to email me!
I'm looking for:
🧑💻 an intern on generative models for change detection
🧑🔬 a PhD student on neurosymbolic generative models for geospatial data
Both starting beginning of 2026.
Details are below, feel free to email me!
November 4, 2025 at 10:08 AM
Want to work on generative models and Earth Observation? 🌍
I'm looking for:
🧑💻 an intern on generative models for change detection
🧑🔬 a PhD student on neurosymbolic generative models for geospatial data
Both starting beginning of 2026.
Details are below, feel free to email me!
I'm looking for:
🧑💻 an intern on generative models for change detection
🧑🔬 a PhD student on neurosymbolic generative models for geospatial data
Both starting beginning of 2026.
Details are below, feel free to email me!
Reposted by David Picard
Apparently it is found under "Author Tasks", thanks @parskatt.bsky.social
November 4, 2025 at 8:47 AM
Apparently it is found under "Author Tasks", thanks @parskatt.bsky.social
Reposted by David Picard
[3/10] Unifying hashing 🔄
Can supervised + unsupervised hashing be done in one framework?
CroVCA aligns binary codes across semantically consistent views:
Augmentations → unsupervised
Class-consistent samples → supervised
🧩 One BCE loss + coding-rate regularizer
Can supervised + unsupervised hashing be done in one framework?
CroVCA aligns binary codes across semantically consistent views:
Augmentations → unsupervised
Class-consistent samples → supervised
🧩 One BCE loss + coding-rate regularizer
November 3, 2025 at 2:30 PM
[3/10] Unifying hashing 🔄
Can supervised + unsupervised hashing be done in one framework?
CroVCA aligns binary codes across semantically consistent views:
Augmentations → unsupervised
Class-consistent samples → supervised
🧩 One BCE loss + coding-rate regularizer
Can supervised + unsupervised hashing be done in one framework?
CroVCA aligns binary codes across semantically consistent views:
Augmentations → unsupervised
Class-consistent samples → supervised
🧩 One BCE loss + coding-rate regularizer
Light on the Seine valley.
I'm at a workshop on AI for cultural heritage organized in the beautiful chateau de Saint Germain en Laye.
(Organisation by GDR IASIS and PEPR ICARE national programs)
I'm at a workshop on AI for cultural heritage organized in the beautiful chateau de Saint Germain en Laye.
(Organisation by GDR IASIS and PEPR ICARE national programs)
November 3, 2025 at 8:30 AM
Light on the Seine valley.
I'm at a workshop on AI for cultural heritage organized in the beautiful chateau de Saint Germain en Laye.
(Organisation by GDR IASIS and PEPR ICARE national programs)
I'm at a workshop on AI for cultural heritage organized in the beautiful chateau de Saint Germain en Laye.
(Organisation by GDR IASIS and PEPR ICARE national programs)
Reposted by David Picard
Looking for a PhD program? It all starts with great supervision. Choose wisely.
www.nature.com/articles/d41...
www.nature.com/articles/d41...
November 1, 2025 at 7:26 PM
Looking for a PhD program? It all starts with great supervision. Choose wisely.
www.nature.com/articles/d41...
www.nature.com/articles/d41...
Reposted by David Picard
Ce genre de chose pourrait être l'occasion aussi d'amener la question de la souveraineté numérique FR par rapport aux USA dans le débat politique, lors des questions au gouvernement par exemple. Reste à trouver un(e) député que ça intéressera ...
La Cour pénale internationale va remplacer l'ensemble de ses outils numériques Microsoft par un logiciel européen (OpenDesk).
Et ce, pour éviter de subir de nouvelles sanctions américaines.
Bonne idée...
L'indépendance numérique devrait être une priorité pour toutes les institutions européennes.
Et ce, pour éviter de subir de nouvelles sanctions américaines.
Bonne idée...
L'indépendance numérique devrait être une priorité pour toutes les institutions européennes.
They should have never become so dependent on Microsoft in the first place.
The #ICC will replace Microsoft’s Office software w the German OpenDesk amid concerns about a new wave of U.S. sanctions that could paralyze the court's day-to-day operations: subscriber.politicopro.com/article/2025...
The #ICC will replace Microsoft’s Office software w the German OpenDesk amid concerns about a new wave of U.S. sanctions that could paralyze the court's day-to-day operations: subscriber.politicopro.com/article/2025...
November 1, 2025 at 3:25 PM
Ce genre de chose pourrait être l'occasion aussi d'amener la question de la souveraineté numérique FR par rapport aux USA dans le débat politique, lors des questions au gouvernement par exemple. Reste à trouver un(e) député que ça intéressera ...
Happy Halloween to all the P(doom)-ers!
🎶
I am the bringer of the end of time for man
I am not here, I am not far away
I am not here, I will eradicate mankind
Into the nothingness from whence they came
🤖
www.youtube.com/watch?v=dnL_...
🎶
I am the bringer of the end of time for man
I am not here, I am not far away
I am not here, I will eradicate mankind
Into the nothingness from whence they came
🤖
www.youtube.com/watch?v=dnL_...
Sentient 6
YouTube video by Nevermore - Topic
www.youtube.com
October 31, 2025 at 11:18 PM
Happy Halloween to all the P(doom)-ers!
🎶
I am the bringer of the end of time for man
I am not here, I am not far away
I am not here, I will eradicate mankind
Into the nothingness from whence they came
🤖
www.youtube.com/watch?v=dnL_...
🎶
I am the bringer of the end of time for man
I am not here, I am not far away
I am not here, I will eradicate mankind
Into the nothingness from whence they came
🤖
www.youtube.com/watch?v=dnL_...
Reposted by David Picard
blog.arxiv.org/2025/10/31/a...
FYI the blog post for the updated policy is out. Our llm future is dire:/
FYI the blog post for the updated policy is out. Our llm future is dire:/
October 31, 2025 at 7:37 PM
blog.arxiv.org/2025/10/31/a...
FYI the blog post for the updated policy is out. Our llm future is dire:/
FYI the blog post for the updated policy is out. Our llm future is dire:/
Reposted by David Picard
I'm super happy about Nicolas' latest work, probably the magnum opus of his PhD.
Read the thread for all the great details.
The main conclusion I draw from this work is that better pretraining, in particular by conditioning on better data, allows us to train SOTA models at a fraction of the cost.
Read the thread for all the great details.
The main conclusion I draw from this work is that better pretraining, in particular by conditioning on better data, allows us to train SOTA models at a fraction of the cost.
We introduce MIRO: a new paradigm for T2I model alignment integrating reward conditioning into pretraining, eliminating the need for separate fine-tuning/RL stages. This single-stage approach offers unprecedented efficiency and control.
- 19x faster convergence ⚡
- 370x less FLOPS than FLUX-dev 📉
- 19x faster convergence ⚡
- 370x less FLOPS than FLUX-dev 📉
October 31, 2025 at 11:39 AM
I'm super happy about Nicolas' latest work, probably the magnum opus of his PhD.
Read the thread for all the great details.
The main conclusion I draw from this work is that better pretraining, in particular by conditioning on better data, allows us to train SOTA models at a fraction of the cost.
Read the thread for all the great details.
The main conclusion I draw from this work is that better pretraining, in particular by conditioning on better data, allows us to train SOTA models at a fraction of the cost.