Sign In

Communications of the ACM

ACM TechNews

Researchers Develop Detection Method to Protect Army Networks


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
Some network connections.

U.S. Army researchers have developed a novel algorithm to protect networks by allowing for the detection of adversarial actions that can be missed by current analytical methods.

Credit: Shutterstock

Researchers at the U.S. Army Combat Capabilities Development Command's Army Research Laboratory, working with colleagues at the University of Notre Dame and Purdue University. have developed a novel algorithm that can protect networks by detecting suspicious activity missed by current analytical methods.

Their goal was to build a higher-order network capable of identifying subtle changes in a stream of data that could indicate suspicious activity.

The researchers developed a high-order network representation, BuildHON+, reflecting real-world phenomena and scales for big data and existing network analysis tools, to perform network analytics.

Said the laboratory's Lance Kaplan, "Using a large-scale synthetic taxi movement data with 11 billion taxi movements, we show how multiple existing anomaly detection methods that depend on first-order network collectively fail to capture anomalous navigation behaviors beyond first-order, and how BuildHON+ can solve the problem."

The study was a collaboration with researchers

From "Researchers Develop Detection Method to Protect Army Networks"

U.S. Army Research Laboratory (08/17/20)
View Full Article

 

Abstracts Copyright © 2020 SmithBucklin, Washington, DC, USA


 

No entries found