SBIR/STTR Award attributes
U.S. Government computer networks have seen an increase in the number and complexity of data breaches over the last few years. The only solution to combat this situation is to incentivize cybersecurity defenders to train more frequently with a purpose to continuously learn new skills regardless of their current skill level and to increase the speed of their Cyber Observe, Orient, Decide and Act (OODA) loop to make it faster than the attacker. The EpiSci and Viasat team propose the Tactical Artificial Intelligence Cybersecurity Training System (TACTS) as a powerful tool for training new and experienced users. TACTS consists of a game built on top of the CORE/EMANE network simulator, with a highly customized set of interactive user interfaces, an intuitive scoring mechanism built on a large collection of Threat Intelligence, a cybersecurity red team agent based on Deep Reinforcement Learning (DRL) for intelligent execution of cyber-attacks without a human-in-the-loop, and a cybersecurity red team agent based on Surprise-Based Learning (SBL) that learns to predict a users behavior and forces them into previously unencountered situations.Approved for Public Release | 18-MDA-9710 (6 Jul 18)
