Current browse context:
stat.ML
Change to browse by:
References & Citations
Statistics > Machine Learning
Title: Global Optimality of Local Search for Low Rank Matrix Recovery
(Submitted on 23 May 2016 (v1), last revised 27 May 2016 (this version, v2))
Abstract: We show that there are no spurious local minima in the non-convex factorized parametrization of low-rank matrix recovery from incoherent linear measurements. With noisy measurements we show all local minima are very close to a global optimum. Together with a curvature bound at saddle points, this yields a polynomial time global convergence guarantee for stochastic gradient descent {\em from random initialization}.
Submission history
From: Srinadh Bhojanapalli [view email][v1] Mon, 23 May 2016 22:05:42 GMT (1060kb,D)
[v2] Fri, 27 May 2016 00:54:17 GMT (1061kb,D)
Link back to: arXiv, form interface, contact.