Phase II Amount
QUINN (Quantum INspired Neural Networks) is a cybersecurity implementation to be deployed alongside a machine learning system (especially a reinforcement learning, active learning, and computer vision system) to harden said system from both standard generative adversarial attacks (noise injection) and cyber-physical attacks. This protection is provided by three unique components that perform attack detection and filtering, identification of attack method and affected aspects, and the cleansing of attack-related features/aspects from the data to recover the original un-altered data instance. These capabilities were informed by a Phase I proof-of-purpose that integrated quantum information theory and classical machine learning protocols to conduct attack filtering with a >86% accuracy (optimized) by exploiting information dense qubit-based ensembles, to leverage a quantum information-based identify modification to data and the fingerprints left by the modification method, and to combine state-of-the-art Generative Adversarial Networks combined with qubit-based information distributions to clean incoming data before it can affect the overall machine learning system. Critically, methods outlined in Phase I conduct these procedures in near-real-time (computational cost of less than a second) in a clandestine manner that cannot be detected, characterized, or circumvented by the adversarial attacker.