SBIR-STTR Award

We propose a quantum adversarial machine learning (QAML) approach that combines ideas from different classical AML techniques such as Defense-GAN and thermometer-encoding of inputs. We propose an implementation of Defense-GAN on the D-Wave Leap quantum computing environment. We also propose to leverage ideas from quantum information science such as noisy inputs/outputs/parameters to improve the robustness of the proposed QAML approach. We will use the MNIST dataset to demonstrate feasibility of the proposed approach.