Hosted by IDTechEx
An expert outlook on the world of AI
HomeEventsReportsTVCareersAbout UsSign-up or LoginIDTechExTwitterFacebookLinkedInGoogle+YoutubeRSSForward To Friend
Posted on June 16, 2017

Artificial neural network to detect computer system faults

Abdullah Muzahid, an assistant professor of computer science at UTSA, has received a $450,000 National Science Foundation (NSF) Faculty Early Career Development award to develop a hardware-based artificial intelligence system that can detect costly software bugs and security attacks in computer systems.
Each year, businesses spend millions on cybersecurity and bug fixes. A 2016 report by the International Data Corporation estimated that more than $73.7 billion is spent worldwide in security-related hardware, software and service expenses. Muzahid and his collaborators hope to significantly decrease those expenditures with their new system.
Over the next five years, the UTSA researcher and his team of undergraduate and graduate students will develop an artificial neural network (ANN) dubbed "NFrame" that can detect, avoid and expose the root causes of system faults, bugs and security attacks. ANNs are computer systems modeled after the human brain and nervous system that are designed to recognize system behaviors and make decisions based on those recognitions.
"Our goal with NFrame is to create a self-policing computer system that is accurate, adaptive and fast," Muzahid said. "Not only is our approach the first to use neural network hardware in this way, but its processes will give new insights into the causes and manifestations of bugs, security flaws and computer system faults."
Webinars Generic Banner
NFrame will monitor correlated code, data and program instructions to learn the "acceptable" or normal behaviors of the various software programs running on its system. Behaviors that deviate from those defined parameters will be identified as bugs or attacks. For example, Muzahid says that
NFrame could tell its users why a specific software keeps crashing, pinpoint a security flaw in a program and report the exact way in which it's compromised, or it could flag and prevent a program from sending information to an unauthorized third-party.
"NFrame can not only tell you why something has gone wrong, but because of how it learns it can also predict when something is about to go wrong in its system," said Muzahid, whose top-tier research focuses on improving the programmability of computer architecture by providing various support in the hardware. "The network can also tell you what is wrong, how it is wrong, where it is wrong, why it is wrong, and whether something will be wrong in the future."
The majority of ANNs are built in software. NFrame will be built into the hardware running alongside its computer systems--allowing it to adapt and evolve with its host-system at incredible speeds. According to Muzahid, hardware-based ANNs are able to process information and make decisions at more than 100 times the speed of software-based networks.
"In an ideal world, we will one day be able to have adaptive artificial neural networks like NFrame on every computer system to help it protect itself from software bugs and other risks that can make it vulnerable to attack or intrusion," Muzahid said.
Source and top image of Abdullah Muzahid: University of Texas at San Antonio