New Talk Online! 🎥 📢 "Unfolding #AlphaFold: How Probability Kinematics Solved the Protein Folding Problem", explores the underrecognized probabilistic foundations of AlphaFold's initial success in 2018 — and shows that its “potentials of mean force” are best understood through

Exciting opportunity: Join DTU Health Tech as an Associate Professor in Deep Learning and Health Bioinformatics. Apply here: ➡️ efzu.fa.em2.oraclecloud.com/hcmUI/Candidat…

🍾 Last year, the Google DeepMind team behind #AlphaFold received the Nobel Prize for solving the protein folding problem, which had been an open problem for more than 50 years. #AlphaFold2, like ChatGPT, relied on the Attention Mechanism to infer which amino acids are close

Xingrui Wang Alan Yuille Wufei Ma Shuo Chen Adam Kortylewski Jack Jingyu Zhang Daniel Khashabi 🕊️ Benjamin Van Durme Ahmed Elgohary Eric Nalisnick’s “ELBOing Stein: Variational Bayes with Stein Mixture Inference” w/ Ola Rønning, Christophe Ley, Padhraic Smyth, and Thomas Hamelryck generalizes Stein variational gradient descent to avoid variance collapse: arxiv.org/abs/2410.22948 (7/12)

Unfolding AlphaFold’s Bayesian Roots in Probability Kinematics １．This theoretical paper reinterprets AlphaFold1 (AF1)’s potential energy function as an instance of probability kinematics (PK), a principled Bayesian update mechanism for incorporating uncertain evidence—offering

🔥 Newly posted on arXiv! "Unfolding AlphaFold’s Bayesian Roots in Probability Kinematics" (with Kanti Mardia) 📄 arxiv.org/abs/2505.19763# We rethink the origins of the Nobel-winning deep learning method that solved the protein folding problem. This paper offers a new view

Sequence Modeling Is Not Evolutionary Reasoning １．This paper makes a bold claim: despite widespread belief, protein language models (PLMs) do not inherently learn evolutionary relationships. Their success on tasks like masked prediction does not imply phylogenetic reasoning

The fourth Philosophy Portal anthology, Writing For (a) First Cause, is now available as both a physical paperback and kindle e-book. For all details, see: philosophyportal.online/writing-for-a-… cc O.G. Rose Lectures on Lacan Thomas Hamelryck @deschoolingmatt

Abstract, Table of Contents, and Order link for the new Philosophy Portal anthology, "Writing For (a) First Cause": philosophyportal.substack.com/p/writing-for-…

🎉Decisions are out for EurIPS 2025 Workshops: eurips.cc/workshops As workshop chairs with yingzhen, @autrechel and Dan Witzner Hansen, we are excited to share the workshop program and look forward to seeing the community in December!

Have you ever asked yourself "What is EurIPS?" or "What is EurIPS short for"? Well luckily #EurIPS have an FAQ addressing some of these questions! And if any questions are unanswered we will do out best to answer them as soon as possible eurips.cc/faq/

How can we make probabilistic ML fast, generalizable, and ready for the real world? 🤔 Join us for the ELLIS UnConference Workshop on Amortized ProbML on Dec 2 in Copenhagen 🇩🇰 co-located with #EurIPS

Reason #1: Was your NeurIPS Conference paper accepted, but you aren't going to San Diego or Mexico City? In Copenhagen, we can offer you great visibility for your paper, in a slightly more intimate space (just you and 1500 other scientists)

Reason #2 When attending the inaugural #EurIPS you will be part of the rising EU AI wave. Together with the ELLIS UnConference you can look forward to engaging with some of the world's most impressive research - all with a European touch.

Go Denmark 😎