Machine Learning Security in Cyber-Physical Systems

Machine learning algorithms are susceptible to adversarial attacks that cause the model to output wrong results by polluting the training data. The problem is exacerbated when the machine learning algorithms are used by critical infrastructures such as the power grid and transportation systems. We demonstrate the insecurity of machine learning applied to these systems by developing realistic adversarial attacks.

Room: 202, Bldg: VIC, Toronto Metropolitan University, Toronot, Ontario, Canada, M5B 2K3