AI Chat AI Image Generator AI Video Text to Speech

Active strict saddles in nonsmooth optimization

12/16/2019

∙

by Damek Davis, et al.

∙

∙

We introduce a geometrically transparent strict saddle property for nonsmooth functions. This property guarantees that simple proximal algorithms on weakly convex problems converge only to local minimizers, when randomly initialized. We argue that the strict saddle property may be a realistic assumption in applications, since it provably holds for generic semi-algebraic optimization problems.

Damek Davis
18 publications
Dmitriy Drusvyatskiy
22 publications

page 4

page 5

research

∙ 02/04/2021

Escaping Saddle Points for Nonsmooth Weakly Convex Functions via Perturbed Proximal Algorithms

We propose perturbed proximal algorithms that can provably escape strict...

0 Minhui Huang, et al. ∙

research

∙ 02/27/2013

Modus Ponens Generating Function in the Class of ^-valuations of Plausibility

We discuss the problem of construction of inference procedures which can...

0 Ildar Z. Batyrshin, et al. ∙

research

∙ 09/13/2022

Semi-strict chordality of digraphs

Chordal graphs are important in algorithmic graph theory. Chordal digrap...

0 Jing Huang, et al. ∙

research

∙ 09/06/2021

Stochastic Subgradient Descent on a Generic Definable Function Converges to a Minimizer

It was previously shown by Davis and Drusvyatskiy that every Clarke crit...

0 Sholom Schechtman, et al. ∙

research

∙ 06/17/2021

Escaping strict saddle points of the Moreau envelope in nonsmooth optimization

Recent work has shown that stochastically perturbed gradient methods can...

0 Damek Davis, et al. ∙

research

∙ 08/04/2021

Stochastic Subgradient Descent Escapes Active Strict Saddles

In non-smooth stochastic optimization, we establish the non-convergence ...

0 Pascal Bianchi, et al. ∙

research

∙ 02/08/2023

A Unified Approach to Unimodality of Gaussian Polynomials

In 2013, Pak and Panova proved the strict unimodality property of q-bino...

0 Christoph Koutschan, et al. ∙