Document Type

Article

Publication Date

7-2017

Publication Title

SIAM Workshop on Dimension Reduction 2017

Conference Name

SIAM Workshop on Dimension Reduction 2017

Conference City

Pittsburgh, PA

Conference Dates

July 9-10, 2017

Abstract

In this work, we present an efficient method for estimating active subspaces using only random observations of gradient vectors. Our method is based on the bi-linear representation of low-rank gradient matrices with a novel initialization step for alternating minimization.

Comments

©SIAM

Share

COinS