Sign In

Communications of the ACM

ACM TechNews

Nomadic Computing Speeds ­p Big Data Analytics


View as: Print Mobile App Share: Send by email Share on reddit Share on StumbleUpon Share on Hacker News Share on Tweeter Share on Facebook
Schematic of the proposed approach to predicting gene-disease associations. First, the researchers construct gene and disease features using different sources. Then, they perform Inductive matrix completion using row and column features. The shaded region

The latest research by University of Texas as Austin professor Inderjit Dhillon is a non-locking, stochastic multi-machine algorithm for asynchronous and decentralized matrix completion called NOMAD.

Credit: Nagarajan Natarajan, Inderjit Dhillon

University of Texas at Austin professor Inderjit Dhillon, a 2014 ACM Fellow, concentrates on expediting big data analytics by using machine learning to reduce data to its most insightful parameters. His latest research, supported by the U.S. National Science Foundation (NSF), is a non-locking, stochastic multi-machine algorithm for asynchronous and decentralized matrix completion (NOMAD).

Dhillon says the algorithm can extract meaningful information from data much faster than other current cutting-edge instruments, as well as investigate data sets that break other leading software.

Issues Dhillon and his collaborators are exploring with NOMAD include topic modeling, in which the system automatically ascertains the appropriate topics related to billions of documents, and recommender systems, in which the system can suggest appropriate items to purchase or people to meet.

Dhillon says NOMAD operates on the principle of distributing computations over different machines using asynchronous communication. "The parameters go to different processors, but instead of synchronizing this computation followed by communication, the nomadic framework does its work whenever a variable is available at a particular processor," Dhillon notes.

NSF program director Amy Apon says the NOMAD approach helps clear a way to run machine algorithms on massive-scale, distributed, commodity systems.

From National Science Foundation
View Full Article

 

Abstracts Copyright © 2015 Information Inc., Bethesda, Maryland, USA


 

No entries found