US 9,870,342 B2 · Approach for more efficient use of computing resources while calculating cross product or its approximation for logistic regression on big data sets

Abstract

According to one technique, a modeling computer computes a Hessian matrix by determining whether an input matrix contains more than a threshold number of dense columns. If so, the modeling computer computes a sparsified version of the input matrix and uses the sparsified matrix to compute the Hessian. Otherwise, the modeling computer identifies which columns are dense and which columns are sparse. The modeling computer then partitions the input matrix by column density and uses sparse matrix format to store the sparse columns and dense matrix format to store the dense columns. The modeling computer then computes component parts which combine to form the Hessian, wherein component parts that rely on dense columns are computed using dense matrix multiplication and component parts that rely on sparse columns are computed using sparse matrix multiplication.

Claims22

Dependent claims

A method comprising:

receiving an input matrix containing a plurality of elements, wherein the elements are organized into one or more rows and one or more columns;
partitioning the one or more columns of the input matrix into one or more sparse columns and one or more dense columns;
identifying a plurality of matrix components that form a cross product matrix, wherein each matrix component of the plurality of matrix components is based on one or more of:
the one or more sparse columns or the one or more dense columns;
computing the cross product matrix, wherein a first set of components of the plurality of matrix components that are based at least partially on the one or more dense columns are computed using dense matrix multiplication and a second set of components of the plurality of matrix components that are based solely on the one or more sparse columns are computed using sparse matrix multiplication;
and
wherein the method is performed by one or more processors.

4 more elements

A non-transitory computer-readable storage medium storing sequences of one or more instructions which, when executed by one or more processors, cause:

receiving an input matrix containing a plurality of elements, wherein the elements are organized into one or more rows and one or more columns;
partitioning the one or more columns of the input matrix into one or more sparse columns and one or more dense columns;
identifying a plurality of matrix components that form a cross product matrix, wherein each matrix component of the plurality of matrix components is based on one or more of:
the one or more sparse columns or the one or more dense columns;
and
computing the cross product matrix, wherein a first set of components of the plurality of matrix components that are based at least partially on the one or more dense columns are computed using dense matrix multiplication and a second set of components of the plurality of matrix components that are based solely on the one or more sparse columns are computed using sparse matrix multiplication.

3 more elements

Dependencies are inferred from the claim text.

Approach for more efficient use of computing resources while calculating cross product or its approximation for logistic regression on big data sets

Life of this patent

Approach for more efficient use of computing resources while calculating cross product or its approximation for logistic regression on big data sets

Life of this patent

Abstract

Claims22