MapReduce Nonnegative Matrix Factorizations (MRNMF)

Zhengguo Sun, Tao Li, and Naphtali Rishe

This code provides a MapReduce implementation of the large-scale nonnegative matrix factorizations described in Large-Scale Matrix Factorization using MapReduce. The implementation uses Hadoop with Java.

This code implements NMF algorithm from paper above.

Usage

With input martix

hadoop jar MRNMF.jar MRNMF -i input.txt -o output -t mrnmf -n 100 -m 20 -k 8 -it 3 -r 5 -a NMF

With random martix

hadoop jar MRNMF.jar MRNMF -s 0.3 -o output -t mrnmf -n 100 -m 20 -k 8 -it 3 -r 5 -a NMF

Options

-h           print help for this application
-i           path to the input matrix file
-s           sparsity of random generated matrix
-o           path to the output directory
-t           path to the temporary directory
-n           matrix row number
-m           matrix column number
-it          iterations number
-k           the dimension number to be reduced to
-r           the range value for the matrix elements
-a           the algorithm to be used

Examples

See Examples for details