PDE Methods in Machine Learning: from Continuum Dynamics to Algorithms (24w5191)


Katy Craig (University of California, Santa Barbara)

Joan Bruna Estratch (New York University)

Lénaïc Chizat (EPFL)

Qin Li (University of Wisconsin-Madison)


The Institute of Mathematics at the University of Granada will host the "PDE Methods in Machine Learning: from Continuum Dynamics to Algorithms" workshop at the University of Granada (IMAG) in Spain, from June 9 - 14, 2024.

Over the past twenty years, machine learning (ML) algorithms have led to breakthroughs in our digital lives. However, theoretical understanding of key methods remains elusive. Traditional approaches from theoretical computer science, based on analyzing algorithms at the fully discrete level, are still far from explaining the mechanisms underlying modern methods.

Instead, the past five years have seen a surge in interest in how the continuum perspective rooted in the study of partial differential equations (PDEs) can shed light on properties of algorithms. At the same time, the new equations arising from ML have attracted interest in the PDE community, as they often fall outside the scope of the existing theory.

Motivated by this interplay, {our workshop will bring together international experts in ML and PDE} to understand how PDE techniques can be used to solve open problems in ML and how problems arising in ML can inspire the development of new mathematical techniques in PDE.

The Banff International Research Station for Mathematical Innovation and Discovery (BIRS) is a collaborative Canada-US-Mexico venture that provides an environment for creative interaction as well as the exchange of ideas, knowledge, and methods within the Mathematical Sciences, with related disciplines and with industry. BIRS is supported by Canada's Natural Science and Engineering Research Council (NSERC), the U.S. National Science Foundation (NSF), Alberta's Advanced Education and Technology, and Mexico's Consejo Nacional de Ciencia y Tecnología (CONACYT).