Gradient Descent Converges to Minimizers

Authors: Jason D. Lee, Max Simchowitz, Michael I. Jordan, Benjamin Recht

Published: 2016 ()

Algorithm: Gradient Descent

arXiv: 1602.04915

Summary

Abstract

We show that gradient descent converges to a local minimizer, almost surely with random initialization. This is proved by applying the Stable Manifold Theorem from dynamical systems theory.

Links

Primary

Paper arxiv.org

Standard

arXiv Abstract 1602.04915 arXiv PDF 1602.04915 arXiv HTML 1602.04915