Brain-Inspired Cybersecurity System Identifies Complex Threats Faster

Sandia National Laboratories’ Roger Suppona, left, and John Naegle, center, and Lewis Rhodes Labs CEO David Follett examine their Neuromorphic Cyber Microscope. This small processor can replace racks of conventional cybersecurity systems. Credit: Randy Montoya

Researchers at Lewis Rhodes Labs in partnership with Sandia National Laboratories have developed the Neuromorphic Cyber Microscope, a practical and revolutionary device that analyzes cyberattacks more than 100 times faster and 1,000 times more energy-efficient than conventional cybersecurity systems.

Advanced cybersecurity systems excel at identifying “bad apples” in computer networks, but they lack the computing power to identify cyber threats directly. As describes, advanced cybersecurity systems “look for general indicators of an attack; call them ‘apples.’ Or the system flags very specific patterns, such as ‘bad Granny Smith apples’ or ‘bad Red Delicious apples.’” This limitation makes it easy for new types of “bad apples” to evade the system, leaving cybersecurity analysts to manually sort the real dangers from false alarms, such as “forbad applesauce”. Neuromorphic Cyber Microscope removes this limitation, by mimicking the neurobiological architecture of the human brain. It is a data streaming processing unit that is faster at analyzing cyberattacks a hundred times more than conventional cybersecurity systems, all while using less energy than a standard 60-watt light bulb.  

The processor in the Neuromorphic Cyber Microscope “breaks data into smaller components and recombines them to make useful information” said computer systems expert John Naegle of Sandia National Laboratories. The processor was developed based on the neuroscience research of Dr. Pamella Follett, a pediatric neurologist, neuroscientist, and co-founder of Lewis Rhodes Labs. Follet studies developmental diseases, such as cerebral palsy in children. Based on her research, David Follett, her husband, co-founder and CEO of Lewis Rhodes Labs, developed a computational model of how the human brain processes information. Brains with cerebral palsy were compared to healthy brains for insights and this led to the development of brain-inspired computer hardware. The Folletts approached Sandia, for its track record of solving real-world problems, where a team led by Naegle sought problems where the neuromorphic processor would make an impact. The team explored robotics and pattern recognition before settling on cybersecurity. “We quickly realized that we could use this architecture to greatly accelerate our ability to look for patterns and even look for complex versions of these patterns,” said Naegle. David Follett notes that the technology is applicable to a wide range of fields including image processing, video processing, speech recognition and behavioral prediction works, among others. “I think if you look out fifteen to twenty years,” he said, “what you’ll see is a device that looks very similar to the graphics processing unit, the GPU, except with a different application, that being the ability to match patterns in various spaces and various complexities.”

The Neuromorphic Cyber Microscope, similar to the human brain, continually scans for threats, so just as a hose or stick can cause you to jump, even if you’re not searching for a snake, the Neuromorphic Cyber Microscope compares streaming data to suspicious patterns in a time-dependent manner. In contrast, conventional cyber detection systems sequentially identify small segments of data against a library of “bad apple” patterns or pre-identified threats. The Neuromorphic Cyber Microscope and a conventional system were tested on Sandia’s cybertraffic in a demo-environment. While the conventional system slowed exponentially as the “bad apples” became more complex, the Neuromorphic Cyber Microscope kept performing efficiently. The Neuromorphic Cyber Microscope is a 2017 R&D100 Awards finalist and is in the early stages of deployment.

No Comments Yet

Leave a Reply

Your email address will not be published.