pith. sign in

arxiv: 1605.00405 · v2 · pith:UWRCZMH6new · submitted 2016-05-02 · 🧮 math.DS · cs.LG

Gradient Descent Only Converges to Minimizers: Non-Isolated Critical Points and Invariant Regions

classification 🧮 math.DS cs.LG
keywords pointsconvergescostcriticaldescentgradientnon-isolatedallowable
0
0 comments X
read the original abstract

Given a non-convex twice differentiable cost function f, we prove that the set of initial conditions so that gradient descent converges to saddle points where \nabla^2 f has at least one strictly negative eigenvalue has (Lebesgue) measure zero, even for cost functions f with non-isolated critical points, answering an open question in [Lee, Simchowitz, Jordan, Recht, COLT2016]. Moreover, this result extends to forward-invariant convex subspaces, allowing for weak (non-globally Lipschitz) smoothness assumptions. Finally, we produce an upper bound on the allowable step-size.

This paper has not been read by Pith yet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Limitations of Lazy Training of Two-layers Neural Networks

    stat.ML 2019-06 unverdicted novelty 8.0

    For quadratic targets in d dimensions, two-layer quadratic networks achieve lower risk when fully trained than in random features or neural tangent regimes if hidden units < d.

  2. Accelerated Gradient Methods for Nonconvex Optimization: Escape Trajectories From Strict Saddle Points and Convergence to Local Minima

    math.OC 2023-07 unverdicted novelty 7.0

    Theoretical analysis of accelerated gradient methods showing almost-sure escape from strict saddles and linear exit times, plus a subclass achieving near-optimal convergence to local minima in convex neighborhoods of ...