Loading…

Faster First-Order Methods for Stochastic Non-Convex Optimization on Riemannian Manifolds

First-order non-convex Riemannian optimization algorithms have gained recent popularity in structured machine learning problems including principal component analysis and low-rank matrix completion. The current paper presents an efficient Riemannian Stochastic Path Integrated Differential EstimatoR...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on pattern analysis and machine intelligence 2021-02, Vol.43 (2), p.459-472
Main Authors: Zhou, Pan, Yuan, Xiao-Tong, Yan, Shuicheng, Feng, Jiashi
Format: Article
Language:English
Subjects:
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:First-order non-convex Riemannian optimization algorithms have gained recent popularity in structured machine learning problems including principal component analysis and low-rank matrix completion. The current paper presents an efficient Riemannian Stochastic Path Integrated Differential EstimatoR (R-SPIDER) algorithm to solve the finite-sum and online Riemannian non-convex minimization problems. At the core of R-SPIDER is a recursive semi-stochastic gradient estimator that can accurately estimate Riemannian gradient under not only exponential mapping and parallel transport, but also general retraction and vector transport operations. Compared with prior Riemannian algorithms, such a recursive gradient estimation mechanism endows R-SPIDER with lower computational cost in first-order oracle complexity. Specifically, for finite-sum problems with n components, R-SPIDER is proved to converge to an 6-approximate stationary point within O(min(n/ε 2 ,1/3)) stochastic gradient evaluations, beating the best-known complexity O(n+1/ ε 4 ); for online optimization, R-SPIDER is shown to converge with O(1/ε 3 ) complexity which is, to the best of our knowledge, the first non-asymptotic result for online Riemannian optimization. For the special case of gradient dominated functions, we further develop a variant of R-SPIDER with improved linear rate of convergence. Extensive experimental results demonstrate the advantage of the proposed algorithms over the state-of-the-art Riemannian non-convex optimization methods.
ISSN:0162-8828
1939-3539
2160-9292
DOI:10.1109/TPAMI.2019.2933841