Luca Pesce

Hi! I am a final-year PhD student at École Polytechnique Fédérale de Lausanne (EPFL) advised by Florent Krzakala working at the intersection of machine learning, statistical physics, and high-dimensional statistics.

My research revolves around understanding the role that data structure plays in the success of learning algorithms, such as Stochatstic Gradient Descent or Bayesian methods. This is motivated by the fact that in the high-dimensional regime, where the number of datapoints and their dimension are growing to infinity together, the presence of low-dimensional relevant subspaces in the data is a crucial ingredient for explaining the performance of modern learning routines. To this end, statistical physics provides a powerful framework to study this problem by formalizing the notion of phase transitions (learning versus not learning) and identifying low-dimensional order parameters (sufficient statistics) that govern the learning process.

Before starting my PhD, I was part of the International Msc in Theoretical Physics organized by a consortium of universities: ICTP, Paris Cité U., Politecnico di Torino, SISSA, Sorbonne U., and U. Paris Saclay.

If you want to know more about my research, the list of my publications is here.