Advanced Comput, Math & Data
Staff Awards & Honors
To Protect and Serve
PNNL's Multiscale Graph Analytics Framework for cyber security featured at GraphLab Conference
During July’s GraphLab Conference 2014, Sutanay Choudhury, a research scientist with Pacific Northwest National Laboratory’s Data Sciences group, hosted a demonstration showcasing M&Ms4Graphs, a graph analytics framework for cyber security. The work stems from the M&Ms4Graphs: Multi-scale, Multi-dimensional Graph Analytics Tools for Cyber Security project, which uses graph-theoretic models to provide continuous updates on system states as part of enabling a resilient (a system’s ability to function in the face of impediments) cyber infrastructure. The project is one of many backed by PNNL’s Asymmetric Resilient Cybersecurity Initiative and features a diverse team of computer scientists and mathematicians from both PNNL’s Fundamental & Computational Sciences and National Security directorates, including major contributors Peter Hui, Kiri Oler, Chase Dowling, Emilie Hogan, Mahantesh Halappanavar, and Sherman Beus.
The web-based interface for the M&Ms4Graphs analytics framework. Enlarge Image.
By studying information flows modeled as large-scale dynamic graphs, Choudhury and his colleagues developed a multiscale framework that can account for behaviors spanning from individual machines to enterprise levels within a cyber system. For the demonstration, Choudhury showed how they have leveraged the GraphLab distributed computing framework to compute a family of cyber security metrics.
The featured visualizations, built as a web application on the Amazon cloud and located here, show the machines in a cyber network. The machines are colored by their behavioral profiles, which are gleaned from the data. In the featured image, the polygon on the right summarizes important properties of the underlying data stream.
At the conference, Choudhury’s presentation received particular interest from the financial security domain, including representatives of a venture capital company and strategists from the Federal Reserve System.
“This attention marks a significant validation of the potential impact for graph-based analysis for cyber security applied to the financial security domain,” Choudhury said.
Held in San Francisco, GraphLab Conference 2014 was the third-annual event that unites experts in graph analytics and large-scale machine learning for a series of presentations, demonstrations, and exhibits. Exhibitors included major companies, such as Oracle Corp., YarcData, Neo Technology Inc. (Neo4j), and MongoDB Inc. Choudhury also showcased this demonstration during the recent Asymmetric Resilient Cybersecurity Initiative annual review held in Richland, Washington.