U.S. Army and industry researchers have developed a metric for neural networks that will measure next generation AI and machine learning (ML) algorithms’ reliability.

“This opens a new research opportunity to create the next generation of algorithms that are robust and resilient,” Dr. Brian Jalaian, a scientist at the U.S. Army Combat Capabilities Development Command’s Army Research Laboratory, said in a press release. “Our approach is versatile and can be added as an additional block to many of the Army’s modern algorithms using modern machine learning algorithms that are based on deep neural networks used for visual imagery.”

Deep neural networks (DNNs) can make predictions once they are trained, but new information that is far outside its training can cause deception. The confidence and reliability of these new metrics will help DNNs learn techniques, apply in command and control systems, precision fire, and decision support systems through safe and secure ML techniques.

“In this work, we proposed a generative model, which adjusts aspects of the original input images in the underlying original deep neural network. The original deep neural network’s response to these generated inputs are then assessed to measure the conformance of the model,” Jalaian said.

Going forward, Army will be looking into applications for the new metric, including cybersecurity and with Internet of Things devices.

Read More About
More Topics
Jordan Smith
Jordan Smith
Jordan Smith is a MeriTalk Senior Technology Reporter covering the intersection of government and technology.