arXiv:2211.14852 Abstract | arXiv Analytics

arXiv:2211.14852 [math.OC]Abstract References Reviews Resources

Sufficient conditions for instability of the subgradient method with constant step size

Published 2022-11-27Version 1

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.

Comments: 14 pages, 3 figures

Categories: math.OC

Keywords: constant step size, subgradient method, sufficient conditions, instability, local minimum

Related articles: Most relevant | Search more

arXiv:2211.14850 [math.OC] (Published 2022-11-27)

Lyapunov stability of the subgradient method with constant step size

Cédric Josz, Lexiao Lai

arXiv:1712.06585 [math.OC] (Published 2017-12-18)

Third-order Smoothness Helps: Even Faster Stochastic Optimization Algorithms for Finding Local Minima

Yaodong Yu, Pan Xu, Quanquan Gu

arXiv:2008.06148 [math.OC] (Published 2020-08-14)

Complexity aspects of local minima and related notions