Building a better (computer) brain for spotting cyberthreats
- By Derek Major
- Feb 29, 2016
Lewis Rhodes Labs was conducting work on neurological diseases when it made a breakthrough that changed its entire business plan.
“We we’re doing neurological modeling to relate autism and several other diseases to the cognitive presentation, and partway through that work all of a sudden the light came on,” David Follett, CEO and co-founder of the West Concord, Mass.-based company, told GCN. “We realized we could use the basic algorithms we developed for medical research to solve a number of pressing problems that exist [for government], cybersecurity being one of them.”
That realization has led Lewis Rhodes Labs to create the Cyber Microscope, a cyber-optimized neuromorphic processor the company claims can increase the speed and resolution of anomaly detection more than 100-fold compared to the average processor.
Follett explained that neuromorphic processors are designed with a unique understanding of the brain and how the sensory cortex processes data. “The brain is very good at pattern matching, and at its first order cybersecurity is about pattern matching,” he said.
Follett said the Cyber Microscope was recently deployed in Sandia National Laboratories' cyber development environment, where the performance gains were tested.
Intrusion detection systems process large amounts of streaming data and alert analysts when an anomaly is detected, but those systems can set off alerts for false positives, making it hard for analysts to find real intruders in their systems.
“The improved speed and accuracy of the LRL Cyber Microscope should allow us to dramatically reduce the false positive rate in our alert database," John Zepper, director of systems mission engineering at Sandia, said in the Lewis Rhodes Labs announcement.
“LRL makes it affordable to deploy Cyber Microscopes virtually everywhere and by doing so, can clearly identify and detect actionable threats,” Follett said.
Derek Major is a former reporter for GCN.