Sufficient conditions for instability of the subgradient method with constant step size
Optimization and Control
2023-06-30 v2
Abstract
We provide sufficient conditions for instability of the subgradient method with constant step size around a local minimum of a locally Lipschitz semi-algebraic function. They are satisfied by several spurious local minima arising in robust principal component analysis and neural networks.
Keywords
Cite
@article{arxiv.2211.14852,
title = {Sufficient conditions for instability of the subgradient method with constant step size},
author = {Cédric Josz and Lexiao Lai},
journal= {arXiv preprint arXiv:2211.14852},
year = {2023}
}
Comments
18 pages, 5 figures