Matthew Nelson
The Intelligence Advanced Research Projects Activity has selected Perspecta's research arm to develop a multiphase mechanism for detecting computer virus that is designed to infiltrate artificial intelligence-based systems.
A Perspecta Labs-led research team it will examine how Trojan works to contaminate the machine learning model training process and create tools to automatically discover the malware in AI platforms, the company said Thursday.
The team aims to establish detection methods that will involve neuron stimulation in an artificial neural network and analyses of adversarial direction clustering and sample input vectors under the agency's TrojAI program.
"We welcome the opportunity to support IARPA on research to defend critical AI systems against malicious interference," said Petros Mouchtaris, president and general manager at Perspecta Labs.
