Ruben Ohana (@oharub) / X

Ruben Ohana

276 posts

Ruben Ohana

@oharub

Senior Research Scientist @NVIDIA | Formerly @FlatironInst, @PolymathicAI & @ENS_ULM | GenAI (for Science) | Views are my own, not my employer's.

San Francisco, CA

Joined April 2013

Pinned
Ruben Ohana
@oharub
Dec 2, 2024
Generating cat videos is nice, but what if you could tackle real scientific problems with the same methods? 🧪🌌 Introducing The Well: 16 datasets (15TB) for Machine Learning, from astrophysics to fluid dynamics and biology. 🐙: github.com/PolymathicAI/t… 📜: openreview.net/pdf?id=00Sx577…
GIF
44K
Ruben Ohana
@oharub
Jul 8, 2025
🧑‍🎓Career update: I recently joined @nvidia as a Research Scientist to work on Foundation Models for synthetic data, with industrial applications! I also moved to SF, didn't expect it to be so different from NYC 😁
42K
Ruben Ohana
@oharub
Jun 27, 2024
🎉 Excited to introduce Gibbs Diffusion (GDiff), a new Bayesian blind denoising method with applications in image denoising and cosmology! #ICML2024 📜 arxiv.org/abs/2402.19455 🔗 github.com/rubenohana/gib… By @HeurtelDepeiges @charlesm993 @oharub @BrunoRegaldo
GIF
GIF
7.9K
Ruben Ohana
@oharub
Oct 4, 2024
Consider joining the Center for Computational Mathematics of the @FlatironInst as a Research Fellow. This is an amazing place for research, in the very heart of NYC, and with a lot of H100 GPUs to play with 😁 Deadline Dec. 15: apply.interfolio.com/155357
7.4K
Ruben Ohana
@oharub
Sep 29, 2021
Proud to announce that out paper "Photonic Differential Privacy with Direct Feedback Alignment" has been accepted at #NeurIPS2021! Happy that NeurIPS reviewers and ACs are open-minded about new hardware for machine learning 😀 A thread will come soon! 📜: arxiv.org/abs/2106.03645
Ruben Ohana
@oharub
Oct 23, 2020
📜 Can we understand Reservoir Computing better? ♋️ Yes! Using kernel methods and Random Features! Our paper accepted as Oral at NeurIPS 2020 with @jonathdong, @MusheghR & @KrzakalaF Paper: arxiv.org/abs/2006.07310 Github: github.com/rubenohana/Res… [1/9]
Ruben Ohana
@oharub
Feb 16, 2022
I am delighted to announce that I accepted the position of Research Fellow at @FlatironCCM starting next October in NYC! I'll collaborate with @gowerrobert and @meickenberg on new optimization methods for deep learning. Exciting times ahead! 🥳
Ruben Ohana
@oharub
Apr 26, 2023
Very excited that our work "Shedding a PAC-Bayesian light on adaptive Sliced-Wasserstein distances" has been accepted at #ICML2023! Joint work with Kimia Nadjahi, @rakotal1 and @LivaRalaivola 🥳
3.3K
Ruben Ohana
@oharub
Dec 18, 2023
Proud of this award, great job from the team!
PolymathicAI
@PolymathicAI
Dec 18, 2023
Very excited to share that our team's "Multiple Physics Pre-training" paper won the Best Paper Award at @AI_for_Science workshop at NeurIPS this year! Congrats to the team and to @mikemccabe210, @BrunoRegaldo @liamhparker and @oharub for leading the effort! 🥳
1.4K
Ruben Ohana
@oharub
Dec 6, 2024
I will be attending #NeurIPS2024! Feel free to ping me or the @PolymathicAI team if you want to chat about The Well 🥤or MPP 🚤! And since you like eye-catchy scientific gifs, here is a Rayleigh-Bénard convection one 😁 🥤: arxiv.org/abs/2412.00568 🚤: arxiv.org/abs/2310.02994
GIF
1.5K
Ruben Ohana
@oharub
Dec 21, 2021
So proud of what the team accomplished these past years ! That’s it, we now have an OPU in a supercomputer 🥳
LightOn
@LightOnIO
Dec 21, 2021
💥This is a world's first: @LightOnIO 's photonic coprocessor is now part of a top500 HPC supercomputer! More in the press release: lighton.ai/news/ lighton.ai/wp-content/upl… Thank you @Genci_fr and #idris @CNRS !
Ruben Ohana
@oharub
Dec 2, 2024
Replying to @oharub
Links: 📜 #NeurIPS2024: openreview.net/pdf?id=00Sx577… 🐙 GitHub: github.com/PolymathicAI/t… 🤗 Blog: huggingface.co/blog/rubenohan… 💻 PyPI: pypi.org/project/the-we… Shout out to the @PolymathicAI team for the great work. What's next is exciting, stay tuned!
848
Ruben Ohana
@oharub
Dec 9, 2020
If you want to learn more about Reservoir Computing (i.e. Recurrent Random Features) and its infinite-width limit (Recurrent Kernels) for chaotic time-series prediction come listen to our *Oral* presentation at #NeurIPS2020 today at 15h15 CET or to Poster Session 4 😁
Ruben Ohana
@oharub
Aug 12, 2024
It was a fun (and difficult) exercise proposed by @neptune_ai to summarize our paper in 100 seconds without any rehearsal. I liked it! Thread about the paper: x.com/oharub/status/…
neptune.ai
@neptune_ai
Aug 11, 2024
Just slightly above the limit, @oharub (Research Fellow at @FlatironCCM) completed his #ICML paper summary in 105.04 secs. Impressive considering the depth of the paper! Title: Listening to the Noise: Blind Denoising with Gibbs Diffusion
00:00
1K