Probabilistic models will be developed for IARPA's insider threat detection program
CAMBRIDGE, Mass. Engineers at Charles River Analytics Inc. will support the Intelligence Advanced Research Projects Activity (IARPA) in the advancement of the Continuous Insider Threat Detection (SCITE) Program. The contract spans over a year and is valued approximately $1.25 million. If all options are exercised in the contract, it could add to the total an additional $2.5 million.
IARPA’s SCITE seeks to develop and test methods for detecting insider threats, which are security risks originating from individuals with malicious intentions within an organization. Charles River Analytics has assembled a team - including Assured Information Security, Applied Marketing Science, and Cognition - to develop models in the Probabilistic Relational Inference Modeling for Enterprises (PRIME) effort for the program.
“In PRIME, we’re developing tools to create and use probabilistic models of inference enterprises that can be used to evaluate and forecast the performance of enterprise-level inference processes, such as detecting personnel conducting industrial espionage in large corporations,” says Dr. Brian Ruttenberg, Senior Scientist at Charles River and Principal Investigator on the PRIME effort. “Using these tools, information technologists, analysts, or other decisions makers in an organization can assess and evaluate inference enterprises and choose the most effective solution for a given domain. PRIME will also provide the capability to propose new and novel inference enterprises, providing effective solutions at a fraction of the cost and with greater assurance than engineered systems.”
Officials say engineers will build probabilistic relational models of inference enterprises in PRIME using the Figaro probabilistic programming language, an open-source language for probabilistic modeling.