Large-Scale Matrix Factorization Using MapReduce
Zhengguo Sun, Tao Li, Naphtali David Rishe · 2010
Due to the popularity of nonnegative matrix factorization and the increasing availability of massive data sets, researchers are facing the problem of factorizing large-scale matrices of dimensions in the orders of millions. Recent research has shown that it is feasible to factorize a million-by-million matrix with billions of nonzero elements on a MapReduce cluster. In this work, we present three different matrix multiplication implementations and scale up three types of nonnegative matrix factorizations on MapReduce. Experiments on both synthetic and real-world datasets show the excellent scalability of our proposed algorithms.