Standard graph neural networks assign vastly different latent embeddings to graphs describing the same physical system at different resolution scales. This precludes consistency in applications and prevents generalization between scales as would fundamentally be needed in many scientific applications. We uncover the underlying obstruction, investigate its origin and show how to overcome it.
The ICLR 2025 MLMP Best Paper Award, along with 2k GPU-hours from Nebius, goes to "On Incorporating Scale into Graph Networks"! Congratulations, Christian Koke, Yuesong Shen,
@abhi-rf.bsky.social, Marvin Eisenberger, @pseudomanifold.topology.rocks, Michael M. Bronstein, @dcremers.bsky.social!
05.05.2025 14:58 β π 6 π 2 π¬ 1 π 1
Equivariant neural networks can effectively model physical systems by naturally handling the underlying geometric quantities and preserving their symmetries, but scaling them to large geometric data remains challenging. Naive downsampling typically disrupts featuresβ transformation laws, limiting their applicability in large scale settings. In this work, we propose a scalable equivariant transformer that efficiently processes geometric data in a coarse-grained latent space while preserving E(3) symmetries of the problem. In particular, by building on the Geometric Algebra Transformer (GATr) and PerceiverIO architectures, our method learns equivariant latent tokens which allow us to decouple the processing complexity from the input data representation while maintaining global equivariance.
ICLR 2025 MLMP best poster award goes to "ViNE-GATr: scaling geometric algebra transformers with virtual nodes embeddings"! Congratulations @sukjulian.bsky.social, Thomas Hehn, @arashbehboodi.bsky.social, Gabriele Cesa!
05.05.2025 14:55 β π 4 π 1 π¬ 1 π 1
π’ Machine Learning Multiscale Processes ICLR 2025 Deadline Extended to Apr 14! π
The good thing it's the Irreproducible Results Track, so the less experiments you run, the greater is the acceptance chance!
Submit here: forms.gle/itcBzomUsLwf...
#ScienceFail #IrreproducibleResults #AcademicHumor
01.04.2025 14:43 β π 0 π 0 π¬ 1 π 0
πContinuing the spotlight series with the next @iclr-conf.bsky.social MLMP 2025 Oral presentation!
πLOGLO-FNO: Efficient Learning of Local and Global Features in Fourier Neural Operators
π· Join us on April 27 at #ICLR2025!
#AI #ML #ICLR #AI4Science
18.03.2025 08:12 β π 2 π 2 π¬ 1 π 0
πContinuing the spotlight series with the next @iclr-conf.bsky.social MLMP 2025 Oral presentation!
π5D Neural Surrogates for Nonlinear Gyrokinetic Simulations of Plasma Turbulence
π· Join us on April 27 at #ICLR2025!
#AI #ML #ICLR
12.03.2025 14:53 β π 0 π 0 π¬ 1 π 0
π Next @iclr-conf.bsky.social 2025 Machine Learning Multiscale Processes Workshop oral presentationπ₯³
πDoMiNO: Down-scaling Molecular Dynamics with Neural Graph Ordinary Differential Equations
π
Join us on April 27 at #ICLR2025!
π· Early reg. deadline: March 15 #AI #ML #ICLR
10.03.2025 07:50 β π 2 π 1 π¬ 1 π 0
Authors: Christian Koke, Yuesong Shen, @abhi-rf.bsky.social, Marvin Eisenberger, @pseudomanifold.topology.rocks, Michael M. Bronstein, @dcremers.bsky.social
08.03.2025 06:49 β π 2 π 1 π¬ 1 π 0
π Excited to kick off our @iclr-conf.bsky.social 2025 Machine Learning Multiscale Processes Workshop contributed paper series! π₯³
π On the Successful Incorporation of Scale into Graph Neural Networks
π
Join us on April 27 at #ICLR2025!
β³ Early reg. deadline: March 15
#AI #ML #ICLR
08.03.2025 06:46 β π 3 π 2 π¬ 1 π 0
ICLR 2025 Workshop MLMP
Welcome to the OpenReview homepage for ICLR 2025 Workshop MLMP
π¨ Paper decisions for the @iclr-conf.bsky.social 2025 Workshop on Machine Learning for Multiscale Processes are out!
π Check out the latest in multiscale ML openreview.net/group?id=ICL...
Save the date: April 27, 2025! ποΈ
Stay tunedβweβll be spotlighting accepted papers soon! π
06.03.2025 11:27 β π 0 π 0 π¬ 0 π 0
π Start writing now, submit on Feb 23! If we solve scale transition, we solve science.
#ICLR2025 #MachineLearning #MultiscaleModeling #AIforScience
13.02.2025 20:04 β π 1 π 0 π¬ 0 π 0
π» GPU grants from our industry partners Constructor Tech and Nebius.
13.02.2025 20:04 β π 0 π 0 π¬ 1 π 0
π βLast callβ to attend @iclr-conf.bsky.social and visit Singapore. Short paper track is specifically designed to be accessible to budding researchers; ICLR offers financial assistance. Singapore is the most expensive city in the world β because it's worth it!
13.02.2025 20:04 β π 0 π 0 π¬ 1 π 0
π¬Top science. The event is part of the β2 AI conference in the world; keynotes by Singapore's only Nobel Laureate, Kostya Novoselov β National University of Singapore, Sergei Gukov β @caltech.edu, Charlotte Bunne β @icepfl.bsky.social, Qianxiao Li β NUS, Daniel Polani β U of Hertfordshire.
13.02.2025 20:04 β π 0 π 0 π¬ 1 π 0
π’ Machine Learning Multiscale Processes @iclr-conf.bsky.social 2025 Deadline Extended to Feb 23! π
Given low-level theory and computationally-expensive simulation code, how can we model complex systems on a useful time scale?
Why attend? π§΅
13.02.2025 20:04 β π 2 π 0 π¬ 1 π 0
Machine Learning Multiscale Processes @ ICLR 2025
Thank you for reading though! We (the multiscale workshop) now have a website, and would like to invite you to update the board) multiscale-ai.github.io
03.02.2025 17:28 β π 1 π 0 π¬ 1 π 0
Machine Learning Multiscale Processes @ ICLR 2025
ππ» Interested in Artificial General Intelligence?
Want to hear more about how Reinforcement Learning could solve Millennium Prize problems?
Attend our workshop and listen to @caltech.edu Prof. Sergei Gukov's talk: Math + AI = AGI π
Submit your work here multiscale-ai.github.io!
28.01.2025 05:53 β π 0 π 0 π¬ 0 π 0
π₯If you think that AI4NA is not a perfect fit for your project, make sure to check out other awesome AI4Science workshops: AgenticAI, @gembioworkshop.bsky.social, @multiscaleai.bsky.social, @climatechangeai.bsky.social, MLGenX
27.01.2025 07:58 β π 2 π 2 π¬ 1 π 0
Machine Learning Multiscale Processes @ ICLR 2025
π₯Call for Submissions for the ICLR2025 @iclr-conf.bsky.social Workshop on ML for Multiscale Processes is openπ₯
Submission deadline is on the 10th of February!
Please submit here: multiscale-ai.github.io
Looking forward to receiving your papers!!!
23.01.2025 09:46 β π 1 π 0 π¬ 0 π 0
Director of Engineering @Qualcomm AI Research
information theory, machine learning, math and philosophy (will post more about it here)
Postdoc @ Technical University of Munich | Intern @ Qualcomm AI Research | PhD @ University of Twente | Neural operators for cardiovascular flow
Professor @ University of Stuttgart, Scientific Advisor @ NEC Labs, GraphML, geometric deep learning, ML for Science and Simulations. Formerly @IUBloomington and @uwcse
ml/nlp phding @ usc, currently visiting harvard, scientisting @ startup;
interpretability & training & reasoning & ai for physics
νamerican, she, (slightly outdated) iglee.me
Postdoc in machine learning with Francis Bach &
@GaelVaroquaux: neural networks, tabular data, uncertainty, active learning, atomistic ML, learning theory.
https://dholzmueller.github.io
AI Pioneer, AI+Science, Professor at Caltech, Former Senior Director of AI at NVIDIA, Former Principal Scientist at AWS AI.
𧬠Structural Bioinformatics | π AI/ML for Drug Discovery | Geometric DL
π¬ @iocbprague.bsky.social, prev. PhD @cusbg.bsky.social @mff.unikarlova.cuni.cz
Co-Founder & Chief Scientist @ Emmi AI. Ass. Prof / Group Lead @jkulinz. Former MSFTResearch, UvA_Amsterdam, CERN, TU_Wien
I like to post papers. They're not always new.
I read a lot: https://dmarx.github.io/papers-feed/
Complexity
Representation Learning
Agnatology, Epistemic Justice
Morality As Cooperation
Ontic Structural Realism
YIMBY, UBI
Research MLE
Former FireFighter
Building generally intelligent agents - ML, RL, Robotics
@nfinite.ai
Professor of Computer Vision and AI at TU Munich, Director of the Munich Center for Machine Learning mcml.ai and of ELLIS Munich ellismunich.ai
cvg.cit.tum.de
Dad Β· Geometry β© Topology β© Machine Learning
Professor at University of Fribourg
While geometry & topology may not save the world, they may well save something that is homotopy-equivalent to it.
π https://bastian.rieck.me/
π« https://aidos.group
Curious Computer Vision Enthusiast w/ @dcremers.bsky.social at @TU Munich
Munich, Germany
π Multilingual enthusiast navigating the realms of English and Esperanto!
π¬ Passionate about science, technology, FOSS and a11y.
π¬ I love engaging in thoughtful discussions
Mastodon: @debby@hear-me.social
Join β: https://hear-me.social/invite/rU3m4Hkc
San Diego Dec 2-7, 25 and Mexico City Nov 30-Dec 5, 25. Comments to this account are not monitored. Please send feedback to townhall@neurips.cc.
International Conference on Learning Representations https://iclr.cc/
PhD Student @CMUPittCompBio.bsky.social / @SCSatCMU.bsky.social
Interested in ML for science/Compuational drug discovery/AI-assisted scientific discovery π€
from π±π°π«Ά https://ramith.fyi
Tackling climate change with machine learning. We facilitate cooperation and provide resources for those working in this area. Share is not endorsement. // https://www.climatechange.ai/
