11/08/2026
Modern AI is full of effects that are easy to observe and hard to explain.
Models can generalize despite having enough capacity to memorize. They can memorize even when trained for generalization. And as models, data, and compute scale, their behavior can shift in ways that intuition alone does not explain.
Lenka Zdeborová looks at learning systems from the level where they start to resemble complex physical systems: many interacting components, collective behavior, and structure that only becomes visible with the right theory.
This is the lens she brings to machine learning, inference, signal processing, and optimization - building solvable models and theoretical principles for understanding how modern AI systems learn, generalize, memorize, and scale.
She’s the next name on our 10th edition lineup.
Lenka Zdeborová is a Professor of Physics and Computer Science at École Polytechnique Fédérale de Lausanne, where she leads the Statistical Physics of Computation Laboratory. She received a PhD in physics from the University of Paris-Sud and Charles University in Prague in 2008, and later spent two years at Los Alamos National Laboratory as the Director's Postdoctoral Fellow. Between 2010 and 2020, she was a researcher at CNRS, working at the Institute of Theoretical Physics in CEA Saclay, France.
Her work has been recognized with the CNRS bronze medal, the Philippe Meyer Prize in theoretical physics, an ERC Starting Grant, the Irène Joliot-Curie Prize, the Gibbs Lectureship of the AMS, the Neuron Fund Award, and, in 2025, an ERC Advanced Grant.
See you in Warsaw for the 10th edition of ML in PL, October 8–10 at the Copernicus Science Centre.
