CUR matrix approximation: Difference between revisions

Revision as of 15:05, 26 June 2012

Revision as of 20:33, 27 June 2012

Line 1:

A CUR matrix approximation is three matrices that, when multiplied together, closely approximate ana given matrix.<ref name=mahoney>{{cite web|title=CUR matrix decompositions for improved data analysis|url=http://www.pnas.org/content/106/3/697.full|accessdate=26 June 2012|author=Michael W. Mahoney|coauthors=Petros Drineas}}</ref> A CUR approximation can be used in the same way as the low-rank approximation of the [[Singular value decomposition]] (SVD). CUR approximations are less accurate than the SVD, but since the rows and columns come from the original matrix (rather than left and right singular vectors), the CUR approximation is often easy for users to comprehend.

Formally, a CUR matrix approximation of a matrix A is three matrices C, U, and R such that C is made from columns of A, R is made from rows of A, and that CUR closely approximates A. Usually the CUR is selected to be a [[Rank (linear algebra)|rank]]-k approximation, which means that C contains k columns of A, R contains k rows of A, and U is a k-by-k matrix. There are many possible CUR matrix approximation, and many CUR matrix approximation of the same rank.