Multiple graph regularized nonnegative matrix factorization

作者：

Highlights：

• Graph model and parameter selection is time consuming and suffers from over fitting.

• We approximate the intrinsic manifold by linear combination of several graphs.

• The graph selection problem is replaced by the solution of multiple graph weights.

• The factorization metrics and the graph weights are learned jointly and iteratively.

摘要

Highlights•Graph model and parameter selection is time consuming and suffers from over fitting.•We approximate the intrinsic manifold by linear combination of several graphs.•The graph selection problem is replaced by the solution of multiple graph weights.•The factorization metrics and the graph weights are learned jointly and iteratively.

论文关键词：Data representation,Nonnegative matrix factorization,Graph Laplacian,Ensemble manifold regularization

论文评审过程：Received 12 June 2012, Revised 11 December 2012, Accepted 5 March 2013, Available online 16 March 2013.

论文官网地址：https://doi.org/10.1016/j.patcog.2013.03.007