Speaker
Domagoj Vlah
(University of Zagreb, Faculty of Electrical Engineering and Computing, Department of Applied Mathematics)
Description
We propose an effective numerical scheme involving deep learning to approximate solution to bilevel optimization problems of size that is considered computationally intractable using known approaches. The lower level is bypassed completely by training a deep neural network to approximate the relevant lower-level effect on the upper level. We illustrate this method on solving bilevel power system optimization problems, commonly arising after the deregulation of the power industry.
This is a joint work with Karlo Šepetanc and Hrvoje Pandžić.
Primary author
Domagoj Vlah
(University of Zagreb, Faculty of Electrical Engineering and Computing, Department of Applied Mathematics)
