A deterministic gradient-based approach to avoid saddle points

DOI10.1017/s0956792522000316MaRDI QIDQ6622959

Stanley J. Osher, Lisa Maria Kreusser, Bei Wang

Publication date: 23 October 2024

Published in: European Journal of Applied Mathematics (Search for Journal in Brave)

zbMATH Keywords

saddle pointsdeterministic algorithmgradient-based methodsattraction regionLaplacian smoothing

Mathematics Subject Classification ID

Nonconvex programming, global optimization (90C26) Numerical optimization and variational techniques (65K10)

Cites Work

• Unnamed Item
• A trust region algorithm with a worst-case iteration complexity of \(\mathcal{O}(\epsilon ^{-3/2})\) for nonconvex optimization
• A geometric analysis of phase retrieval
• Exploiting negative curvature in deterministic and stochastic optimization
• First-order methods almost always avoid strict saddle points
• Cubic regularization of Newton method and its global performance
• Learning Deep Architectures for AI
• A Newton-Based Method for Nonconvex Optimization with Fast Evasion of Saddle Points
• Finding approximate local minima faster than gradient descent
• Gradient Descent Finds the Cubic-Regularized Nonconvex Newton Step
• Learning representations by back-propagating errors
• Laplacian Smoothing Stochastic Gradient Markov Chain Monte Carlo

This page was built for publication: A deterministic gradient-based approach to avoid saddle points