ScalableNMF is
the implementation of Scalable Nonnegative Matrix Factorization
introduced in the paper Scalable
Nonnegative Matrix Factorization with Block-wise Updates,
(Jiangtao Yin, Lixin Gao, and Zhongfei Zhang) in ECML/PKDD 2014.
ScalableNMF supports Nonnegative Matrix Factorization at a million-by-million scale. It is implemented in Java for the Hadoop platform and is free to distribute for non-commercial usage.
A
Hadoop cluster should be running, configured in either
pseudo-distributed or cluster mode. To deploy a Hadoop
cluster
environment, you can refer to Hadoop
cluster setup instructions,
if you've never used Hadoop (we recommend Hadoop 0.20.2 or above for
compatibility).
Download the package, ScalableNMF.tar.gz,
See the README file inside the package for usage instructions.