Accession Number : AD1033864

Title :   Enforced Sparse Non-Negative Matrix Factorization

Descriptive Note : Technical Report

Corporate Author : MIT Lincoln Laboratory Lexington United States

Personal Author(s) : Gadepally,Vijay ; Kepner,Jeremy ; Gavin,Brendan

Full Text :

Report Date : 23 Jan 2016

Pagination or Media Count : 10

Abstract : Non-negative matrix factorization (NMF) is a dimensionality reduction algorithm for data that can be represented as an undirected bipartite graph. It has become a common method for generating topic models of text data because it is known to produce good results, despite its relative simplicity of implementation and ease of computation. One challenge with applying the NMF to large datasets is that intermediate matrix products often become dense, thus stressing the memory and compute elements of the underlying system. In this article, we investigate a simple but powerful modification of the alternating least squares method of determining the NMF of a sparse matrix that enforces the generation of sparse intermediate and output matrices. This method enables the application of NMF to large datasets through improved memory and compute performance. Further, we demonstrate, empirically, that this method of enforcing sparsity in the NMF either preserves or improves both the accuracy of the resulting topic model and the convergence rate of the underlying algorithm.

Descriptors :   algorithms , text mining , sparse matrix , heuristic methods , digital data , natural language computing , bayesian networks , clustering , linear algebra , machine learning , natural languages

Subject Categories : Linguistics

Distribution Statement : APPROVED FOR PUBLIC RELEASE