Numerical methods for optimization problems arising in ML
Title: Numerical methods for optimization problems arising in ML
SNIC Project: Berzelius-2022-155
Project Type: LiU Berzelius
Principal Investigator: Måns Williamson <mans.williamson@math.lth.se>
Affiliation: Lunds universitet
Duration: 2022-08-08 – 2023-03-01
Classification: 10105
Homepage: https://portal.research.lu.se/sv/persons/m%C3%A5ns-williamson
Keywords:

Abstract

The project aims to investigate how numerical methods for solving differential equations can be applied to large scale optimization problems (e.g. training neural networks). An example of this is the stochastic gradient descent algorithm; this can be viewed as a stochastic version of the explicit Euler scheme applied to the gradient flow equation.