Reposted by Antoine WehenkelGilles Louppe @glouppe.bsky.social · 24/09/2026Finally some good news today, Omer's last paper is accepted at NeurIPS! Well done @orochman.bsky.social :-) 182
Reposted by Antoine WehenkelFrançois-Xavier Briol @fxbriol.bsky.social · 28/08/2025Just finished delivering a course on 'Robust and scalable simulation-based inference (SBI)' at Greek Stochastics. This covered an introduction to SBI, open challenges, and some recent contributions from my own group. The slides are now available here: fxbriol.github.io/pdfs/slides-.... 1359
Antoine Wehenkel @awehwe.bsky.social · 11/07/2025Very excited about this 👇 Stay tuned for the call for papers and more info!! ☄️ 030
Reposted by Antoine WehenkelGaspard Lambrechts @gsprd.be · 13/06/2025Two months after my PhD defense on RL in POMDP, I finally uploaded the final version of my thesis :) You can find it here: hdl.handle.net/2268/328700 (manuscript and slides). Many thanks to my advisors and to the jury members. 082
Reposted by Antoine WehenkelGaspard Lambrechts @gsprd.be · 09/06/2025📝 Our paper "A Theoretical Justification for Asymmetric Actor-Critic Algorithms" was accepted at #ICML! Never heard of "asymmetric actor-critic" algorithms? Yet, many successful #RL applications use them (see image). But these algorithms are not fully understood. Below, we provide some insights. 1177
Reposted by Antoine WehenkelGilles Louppe @glouppe.bsky.social · 04/06/2025Positions remain open! Both PhD and postdoctoral opportunities are available on scientific foundation models. An additional position is also available on AI for regional climate models (jointly with @xavierfettweis.bsky.social). Do not hesitate to apply! 0158
Antoine Wehenkel @awehwe.bsky.social · 02/06/2025Huge thanks to my co-authors 🙏 Juan L. Gamella, Ozan Sener, Jens Behrmann, Guillermo Sapiro, Jörn Jacobsen, Marco Cuturi. We hope RoPE helps reframe model misspecification as a learning problem that requires real-world data to be solved. 000
Antoine Wehenkel @awehwe.bsky.social · 02/06/2025This is just the beginning. We hope RoPE pushes SBI toward: ✅ Embracing real-world constraints ✅ Blending domain knowledge + data ✅ Treating robust inference as a learning problem whose objective must be tight with the downstream application for the result of this inference 110
Antoine Wehenkel @awehwe.bsky.social · 02/06/2025Here’s what RoPE does: 1️⃣ Uses a small calibration set of real (x, θ) pairs 2️⃣ Learns a correction from simulated to real obs using optimal transport 3️⃣ Enables simulation-based inference you can actually trust 110
Antoine Wehenkel @awehwe.bsky.social · 02/06/2025That’s what motivated RoPE, our method being presented at ICML 2025 🎉 RoPE reframes misspecification as a posterior inaccuracy problem, not a simulator/data mismatch in contrast to how model misspecification is often defined in the literature. 110
Antoine Wehenkel @awehwe.bsky.social · 02/06/2025Humans design robust statistics by intuition. Neural SBI doesn’t — unless we teach it how. 🔑 Insight: To make SBI robust, show it real-world data. And use labeled data to trust the newly learnt inference pipeline. 120
Antoine Wehenkel @awehwe.bsky.social · 02/06/2025Why? Because inference method and simulator misspecification are deeply entangled. 🧠 Neural SBI often overfits to quirks in simulators. 🤔 Simpler methods (like ABC with handcrafted stats) often perform better when simulators are slightly wrong. 110
Antoine Wehenkel @awehwe.bsky.social · 02/06/2025SBI thrives in ideal settings — but what happens when simulators aren’t perfect? Real-world practitioners always ask: “But what if the simulator is off?” I used to think this was an issue related to the simulator and not to SBI. Now I believe this is the central issue with existing SBI algorithms. 110
Antoine Wehenkel @awehwe.bsky.social · 02/06/2025📣 New paper alert — To be presented at ICML 2025! arxiv.org/abs/2405.08719 What does it really mean for a simulator to be misspecified, if our goal is to estimate parameters with calibrated uncertainty? A 🧵on our new method, RoPE, and what it means for real-world SBI ⬇️arxiv.orgAddressing Misspecification in Simulation-based Inference through Data-driven CalibrationDriven by steady progress in deep generative modeling, simulation-based inference (SBI) has emerged as the workhorse for inferring the parameters of stochastic simulators. However, recent work has dem... 1184
Reposted by Antoine WehenkelFrançois Rozet @francois-rozet.bsky.social · 23/05/2025📢 I am looking for AI post-doc/research/engineer positions in Europe (Paris, London, Zurich, ...) starting 2026. My work revolves around generative modeling and AI for Science, with 4+ publications at top conferences during my PhD. If you are hiring, please reach out! If not, please repost 🔁 21911
Reposted by Antoine WehenkelGilles Louppe @glouppe.bsky.social · 23/05/2025I can't recommend @francois-rozet.bsky.social enough👇He is both this excellent researcher and coder that any ML team would dream having onboard CC: @danilojrezende.bsky.social @awehwe.bsky.social @bkmi.bsky.social @yann-lecun.bsky.social @johannbrehmer.bsky.social 093
Reposted by Antoine WehenkelMackenzie Weygandt Mathis @trackingactions.bsky.social · 22/05/2025But if my lab can help, please DM me. 192
Reposted by Antoine WehenkelarXiv cs.LG Machine Learning @cslg-bot.bsky.social · 29/04/2025Andry, Rozet, Lewin, Rochman, Mangeleer, Pirlet, Faulx, Gr\'egoire, Louppe: Appa: Bending Weather Dynamics with Latent Diffusion Models for Global Data Assimilation arxiv.org/abs/2504.18720 arxiv.org/pdf/2504.18720 arxiv.org/html/2504.18720 146
Reposted by Antoine WehenkelSaganism @sagan.bsky.social · 30/04/2025This is from The Tonight Show with Johnny Carson aired on May 20th, 1977. Carl Sagan says something very important, a strong message that didn't lose any validity since then. 411148344057
Reposted by Antoine WehenkelSander Dieleman @sedielem.bsky.social · 15/04/2025New blog post: let's talk about latents! sander.ai/2025/04/15/l...sander.aiGenerative modelling in latent spaceLatent representations for generative models. 37518
Reposted by Antoine WehenkelMarvin Schmitt @marvin-schmitt.com · 31/03/2025Fantastic initiative by @serge.belongie.com and Søren Hauberg 🇪🇺🇩🇰 Please take a moment to answer through the poll below and share in your networks! 👇 0195
Reposted by Antoine WehenkelMarvin Schmitt @marvin-schmitt.com · 28/03/2025AI is just a hoax by big IKEA to sell more bedrooms. 061
Reposted by Antoine WehenkelThomas Dietterich @tdietterich.bsky.social · 18/03/2025This remains important today. As industry places huge financial bets on large language models, the research community needs to investigate other approaches that may achieve similar or better results via entirely different methods. end/ 1164
Reposted by Antoine WehenkelGilles Louppe @glouppe.bsky.social · 11/03/2025The two postdoctoral positions remain open! (the PhD position has been filled) Applications from deep learning researchers or atmospheric scientists are welcome! Ping me if you have any questions! 0127
Reposted by Antoine WehenkelML for Science @ml4science.bsky.social · 06/03/2025Great work led by @maximiliandax.bsky.social now out in @nature.com: Neural network deciphers gravitational waves from merging neutron stars in a second - new method could be instrumental in preparing the field for the next generation of observatories: is.mpg.de/news/neural-... #AIforScience 1/2is.mpg.deNeural network deciphers gravitational waves from merging neutron stars in a second | Max Planck Institute for Intelligent SystemsBinary neutron star mergers emit gravitational waves followed by light. To fully exploit these observations and avoid missing key signals, speed is crucial. In a study to be published in Nature on Mar... 1125
Reposted by Antoine WehenkelGilles Louppe @glouppe.bsky.social · 05/03/2025Now accepted at @tmlr-pub.bsky.social 🥳 2303
Reposted by Antoine WehenkelSam Power @spmontecarlo.bsky.social · 20/02/2025Reading the recent literature on "neural samplers" is an odd experience. There's a lot of attention drifting towards it (conceivably prompted by the increased attention on diffusion models and the like), and so many people are trying many things. It's a different way of thinking about things to me. 1141
Reposted by Antoine WehenkelGilles Louppe @glouppe.bsky.social · 06/02/2025Old is new. It's funny how so many ideas from what some call "good old-fashioned AI" keep resurfacing! My advice: revisit Russel and Norvig's book with 2025 deep learning. 0131
Reposted by Antoine WehenkelGilles Louppe @glouppe.bsky.social · 06/02/2025I am at the AI Action Summit today and tomorrow! Ping me if you are around! @polytechniqueparis.bsky.social 1161
Reposted by Antoine WehenkelMarco Cuturi @marcocuturi.bsky.social · 22/01/2025Today is a great day for optimal transport 🎉! Lots of gratitude 🙏 for all folks who contributed to ott-jax.readthedocs.io and pushed for the MOSCOT (now @ nature!) paper, from visionaries @dominik1klein.bsky.social, G. Palla, Z. Piran to the magician, Michal Klein! ❤️ www.nature.com/articles/s41...nature.comMapping cells through time and space with moscot - NatureMoscot is an optimal transport approach that overcomes current limitations of similar methods to enable multimodal, scalable and consistent single-cell analyses of datasets across spatial and temporal... 0227
Reposted by Antoine WehenkelJakob Macke @jakhmack.bsky.social · 20/01/2025SBI hackathon is back! Join us in Tübingen for the 4th edition, make the toolbox better, brainstorm new direction, have fun! 0135
Reposted by Antoine WehenkelGilles Louppe @glouppe.bsky.social · 17/01/2025What channels would you recommend for sharing PhD/post-doc position offers? The theme is foundation models for the physical sciences, with weather as a specific target. 3145
Reposted by Antoine WehenkelJes Frellsen @jesfrellsen.bsky.social · 10/01/2025Excited about 🎓 Generative Modeling Summer School / Statlearn 🗓️ Mar 31 - Apr 4, 2025, France ℹ️ gemss.ai 🎟️ Deadline: Jan 27, 2025 🧑🏫 Yingzhen Li, @stephanmandt.bsky.social, Aude Sportisse, Anna Korba, @glouppe.bsky.social, @jesfrellsen.bsky.social, @pamattei.bsky.social, @jmtomczak.bsky.social, TBAgemss.aiGenerative Modeling Summer School (GeMSS) 02511
Reposted by Antoine WehenkelJanne Lappalainen @lappalainenjk.bsky.social · 30/12/2024Ever wanted to do deep learning with a neural net that is one-to-one mapped to 65.05% of the fruit fly brain? 😄 Before this year ends, I will quickly advertise our code release of `flyvis.` Docs: t.ly/YqWzR Repo: t.ly/pMWpp Work with @jakhmack.bsky.social, @srinituraga.bsky.social and colleagues 14817
Antoine Wehenkel @awehwe.bsky.social · 30/12/2024The maxim “Knowledge is power” appears much more profound after reading this really cool blogpost 🤯 www.quantamagazine.org/what-is-entr...quantamagazine.orgWhat Is Entropy? A Measure of Just How Little We Really Know. | Quanta MagazineExactly 200 years ago, a French engineer introduced an idea that would quantify the universe’s inexorable slide into decay. But entropy, as it’s currently understood, is less a fact about the world th... 020
Antoine Wehenkel @awehwe.bsky.social · 20/12/2024Congratulations!! Seems like a nice and well deserved Xmas gift! 010
Reposted by Antoine WehenkelJoey Bose @joeybose.bsky.social · 18/12/2024🔊 Super excited to announce the first ever Frontiers of Probabilistic Inference: Learning meets Sampling workshop at #ICLR2025 @iclr-conf.bsky.social! 🔗 website: sites.google.com/view/fpiwork... 🔥 Call for papers: sites.google.com/view/fpiwork... more details in thread below👇 🧵 28519
Antoine Wehenkel @awehwe.bsky.social · 18/12/2024Astonished at how many professional scientists do use parameter grids 🙃 021
Reposted by Antoine WehenkelAndrew Gordon Wilson @andrewgwils.bsky.social · 18/12/2024This is your monthly reminder that understanding deep learning does not require rethinking generalization, and it never did. 3223
Reposted by Antoine WehenkelKyle Cranmer @kylecranmer.bsky.social · 16/12/2024Got a photo with a nice branch of my academic family tree: @awehwe.bsky.social @glouppe.bsky.social @smsharma.bsky.social @lukasheinrich.com Nicole Hartman @annalenakofler.bsky.social 1305
Reposted by Antoine WehenkelArnaud Doucet @arnauddoucet.bsky.social · 15/12/2024The slides of my NeurIPS lecture "From Diffusion Models to Schrödinger Bridges - Generative Modeling meets Optimal Transport" can be found here drive.google.com/file/d/1eLa3...drive.google.comBreimanLectureNeurIPS2024_Doucet.pdf 932868
Reposted by Antoine WehenkelLukas Heinrich @lukasheinrich.com · 15/12/2024Proud of Nicole Hartman running the show at the ML for Physical Sciences Workshop at NeurIPS happening today 0274
Reposted by Antoine WehenkelKyle Cranmer @kylecranmer.bsky.social · 11/12/2024I love that this post has more than 2 times more likes / reposts than the same one on X despite having more than 4x the number of followers on X 1372
Reposted by Antoine WehenkelKyle Cranmer @kylecranmer.bsky.social · 10/12/2024The papers and posters for our Machine Learning and Physical Sciences workshop at #NeurIPS2024 are online #ml4ps2024. Come check it out on Sunday ml4physicalsciences.github.io/2024/ 011119