Some recent talks
A selection of some relatively recent (\(\lesssim\) 1 year) talks -- not including blackboard talks. This selection is not exhaustive: a complete list of my talks is instead available in my CV. If not provided, feel free to contact me regarding the material of one of my talks.
Learning from long sequences of high-dimensional tokens, and extensive-width neural networks
February 06, 2025
Workshop "Towards a theory of typical-case algorithmic hardness", at Les Houches.
Some results on average-case matrix discrepancy
January 28, 2025
Spin glasses and related topics day, at the Inhomogeneous Random Systems annual workshop at Institut Henri Poincaré.
From fitting ellipsoids to random points, to learning in large neural networks (new version)
September 04, 2024
Roccella Conference on Inference and AI. Extended version of the LemanTH talk below, with more details on the neural network theory part and less on ellipsoid fitting.