DistributedMemory Parallel Symmetric Nonnegative Matrix Factorization
Abstract
We develop the first distributedmemory parallel implementation of Symmetric Nonnegative Matrix Factorization (SymNMF), a key data analytics kernel for clustering and dimensionality reduction. Our implementation includes two different algorithms for SymNMF, which give comparable results in terms of time and accuracy. The first algorithm is a parallelization of an existing sequential approach that uses solvers for non symmetric NMF. The second algorithm is a novel approach based on the GaussNewton method. It exploits secondorder information without incurring large computational and memory costs. We evaluate the scalability of our algorithms on the Summit system at Oak Ridge National Laboratory, scaling up to 128 nodes (4,096 cores) with 70% efficiency. Additionally, we demonstrate our software on an image segmentation task.
 Authors:

 Georgia Institute of Technology
 Wake Forest University, WinstonSalem
 ORNL
 Georgia Institute of Technology, Atlanta
 Publication Date:
 Research Org.:
 Oak Ridge National Lab. (ORNL), Oak Ridge, TN (United States)
 Sponsoring Org.:
 USDOE
 OSTI Identifier:
 1798617
 DOE Contract Number:
 AC0500OR22725
 Resource Type:
 Conference
 Resource Relation:
 Conference: The International Conference for High Performance Computing, Networking, Storage and Analysis  Atlanta, Georgia, United States of America  11/16/2020 10:00:00 AM11/19/2020 10:00:00 AM
 Country of Publication:
 United States
 Language:
 English
Citation Formats
Eswar, Srinivas, Hayashi, Koby, Ballard, Grey, Kannan, Ramakrishnan, Vuduc, Richard, and Park, Haesun. DistributedMemory Parallel Symmetric Nonnegative Matrix Factorization. United States: N. p., 2020.
Web.
Eswar, Srinivas, Hayashi, Koby, Ballard, Grey, Kannan, Ramakrishnan, Vuduc, Richard, & Park, Haesun. DistributedMemory Parallel Symmetric Nonnegative Matrix Factorization. United States.
Eswar, Srinivas, Hayashi, Koby, Ballard, Grey, Kannan, Ramakrishnan, Vuduc, Richard, and Park, Haesun. 2020.
"DistributedMemory Parallel Symmetric Nonnegative Matrix Factorization". United States. https://www.osti.gov/servlets/purl/1798617.
@article{osti_1798617,
title = {DistributedMemory Parallel Symmetric Nonnegative Matrix Factorization},
author = {Eswar, Srinivas and Hayashi, Koby and Ballard, Grey and Kannan, Ramakrishnan and Vuduc, Richard and Park, Haesun},
abstractNote = {We develop the first distributedmemory parallel implementation of Symmetric Nonnegative Matrix Factorization (SymNMF), a key data analytics kernel for clustering and dimensionality reduction. Our implementation includes two different algorithms for SymNMF, which give comparable results in terms of time and accuracy. The first algorithm is a parallelization of an existing sequential approach that uses solvers for non symmetric NMF. The second algorithm is a novel approach based on the GaussNewton method. It exploits secondorder information without incurring large computational and memory costs. We evaluate the scalability of our algorithms on the Summit system at Oak Ridge National Laboratory, scaling up to 128 nodes (4,096 cores) with 70% efficiency. Additionally, we demonstrate our software on an image segmentation task.},
doi = {},
url = {https://www.osti.gov/biblio/1798617},
journal = {},
number = ,
volume = ,
place = {United States},
year = {2020},
month = {11}
}