Massimiliano Pontil

Proud of our contributions to this year's NeurIPS conference

At #NeurIPS2024 🇨🇦 our group will present 7 contributions! These span a diverse array of topics: from theoretical advances in stochastic processes and reinforcement learning to applications in molecular dynamics and uncertainty quantification.

@riccardograzzi @julien_siems @jkhfranke @ZelaArber @FrankRHutter Check out our work led by @riccardograzzi @IIT, in collaboration with @julien_siems @jkhfranke @ZelaArber and @FrankRHutter @UniFreiburg on improving expressivity of LLMs. Linear RNNs are elegant and could be a game changer for LLMs!

LLMs can now track states, finally matching this cat! And we prove it. But how? 🧵👇 1/ Paper: arxiv.org/abs/2411.12537 with @julien_siems @jkhfranke @ZelaArber  @FrankRHutter   @MPontil

@EmtiyazKhan @Korba_Anna @MIJUNGPARK11 @gabrielpeyre @btreetaiji and Philippe Rigollet, Bernhard Schmitzer, @MPontil, @thjashin, Claudia Totzeck, Oliver Tse, Chaoyue Zhao. Excited to have these brilliant minds on board! #OTBerlin2024

Exited so see interest in our recent work. Please visit our project website after the holidays, as we will release all code and examples. danfoa.github.io/DynamicsHarmon…

Heya, our team from @IITalk got the second place in this year’s @OpenCatalyst challenge!

Currently presenting MEKRR at poster #107! Come chat about GNNs, mean embeddings and physics!

@pie_novelli This paper derives the first eigenvector/value learning guarantees for Transfer/Koopman operator estimators. Highly relevant for data-driven dynamical systems! Check pur paper and poster @NeurIPSConf

Learning rates for eigenvalues?! In arxiv.org/abs/2302.02004 we study how well one can learn the spectrum of an operator from data, with direct implications on learning dynamical systems via Koopman / Transfer operator.

@NicoloCB congrats and well deserved, nicolo'! there are now 2 bottles of wine waiting for you!

Proud new member of the Accademia dei Lincei (applied math)! One of the oldest and most prestigious European scientific institutions.

Congratulations to Arkadi Nemirovski and Yurii Nesterov that have been awarded the World Laureates Association Prize, a new international science prize established in 2021. Last year the prize for the same category went to Michael Jordan. thewlaprize.org h/t @GeorgeLan7

The #ELLISPhD Program is open to candidates from around the world! Apply by November 15 if you are interested in doing a #PhD with top #machinelearning labs in Europe: ellis.eu/news/ellis-phd… #AI #ML #PhDProgram

Today, there was a fascinating workshop on Technical Robustness& Safety on AI led by @biggiobattista. Battista is Associate Professor at the @univca and co-founder of the cybersecurity company @pluribus_one . We are grateful for our various partners in the ELSA project.

Just concluded an exhilarating two-day kick-off for #ELIASProject 🤝 Exciting times ahead as we explore #SustainableAI #TrustworthyAI & Nurturing Excellence in #AI over the next four years. 🎯 Our mission: Drive innovation, foster cohesion and promote trustworthy AI! Stay tuned!

Final abstract submission deadline (talks and posters): June 11th, 2023 bilevelconference2023.org International Conference on Bilevel Optimization Southampton, UK, 8-11 August 2023

COMPUTATIONAL POSTDOCS IN MY LAB at IIT @IITalk - 2 OPENINGS. MD simulations, nucleic acids, drug design, ML - in collab with experimentalists. Please reach out for more info. DM to me, or send me an email with CV and a cover letter! - thanks !!! Links available soon.

A roadmap for Europe: a statement on the need for computational resources in Europe, urging to establish an intergovernmental, multi-centric AI research organization to secure Europe’s sovereignty in #AI. ellis.eu/news/ai-founda…

Interested in bilevel opt? See our recent paper, co-authored with Riccardo Grazzi and Saverio Salzo, covering applications to HPO, meta-learning, data poisoning attacks, and more: jmlr.org/papers/volume2… Our algorithm offers optimal convergence without the need for warm restart.

It is a great honor to be part of this amazing lineup of speakers! I look forward to attending this workshop!

Thank you for the interesting discussions! We had a great exchange about the development and deployment of state of the art AI solutions, especially with regard to data privacy and infrastructures. We are looking forward to the next workshop and to advance the vision of ELSA.