Accelerated optimization algorithms: the good, the bad and the odd

Title

Accelerated optimization algorithms: the good, the bad and the odd

Speaker

Juan Peypouquet - University of Groningen

Abstract

We give an accessible introduction to accelerated optimization algorithms. We will comment on the main ideas--including physical principles--that have led to a deeper understanding of their behavior, and produced new design strategies to further improve their numerical performance.

Bio

Juan Peypouquet obtained his PhD from Sorbonne Université (formerly Pierre et Marie Curie - Paris VI) and the University of Chile in 2007, and his Habilitation from Sorbonne Université in 2014. He studies the convergence and complexity of first order numerical optimization methods, including accelerated variants and extensions to equilibrium and fixed point problems. He is currently the chair of Optimization at the Bernoulli Institute for Mathematics, Computer Science and Artificial Intelligence of the University of Groningen in the Netherlands.

< br>

When

Thursday February 8th, 12:00

Where

Room 322 @ DIBRIS/DIMA, Via Dodecaneso 35, Genoa