Online Program Home
My Program

Abstract Details

Activity Number: 449
Type: Contributed
Date/Time: Tuesday, August 2, 2016 : 2:00 PM to 2:45 PM
Sponsor: Section on Statistical Learning and Data Science
Abstract #321569
Title: Likelihood Methods for Non-Negative Matrix Factorization
Author(s): Frank Shen*
Companies: Penn State University
Keywords: Nonnegative matrix factorization ; Dimension Reduction ; Clustering ; Hi-C ; high dimensional data ; chromatin conformation

Nonnegative matrix factorization (NMF) is a machine learning technique which can be used for dimension reduction and unsupervised soft clustering. It became popular, because it is supposed to provide sparse decomposition without an explicit sparsity-based penalty term. Most NMF formulations are only loosely associated with a likelihood based model. However, we have developed a likelihood-based NMF method. By providing an explicit probability model, it is easier to allow for multiple layers of variation in the data. It also allows NMF to be used to test reproducibility in high-dimensional data. Such tests are especially useful for noisy high-dimensional biological data, such as chromatin conformation data.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2016 program

Copyright © American Statistical Association