Ò»±¾µÀÎÞÂë

Cyber-physical smart infrastructures, such as smart grids, buildings, transportation, and pipeline networks, are becoming universal. These cyber-physical systems (CPS) face a wide range of emerging threats originating from both natural and adversarial sources. Decision-makers increasingly need the next-generation modeling and estimation algorithms to enhance the resiliency of CPS. However, existing approaches often lack robustness, leading to inadequate system resiliency. My research addresses these challenges in two sequential parts. In the first part, I explore how sparsity enables robust system modeling to improve resilience against natural random threats. Then, in the second part, I explore how ML can further enhance robust system modeling against modern adversarial threats, such as cyberattacks, through a novel framework termed “Physics-ML Synergy.” These insights have led to new strategies for estimating system states, threats, sources of failure, and corrective actions. Finally, I will discuss the significant potential of Physics-ML Synergy in advancing system defense, planning, and its applications across various domains for a smart, resilient, and sustainable future.

Shimiao (Cindy) Li is a final-year PhD candidate in Electrical and Computer Engineering (ECE) at Ò»±¾µÀÎÞÂë, advised by Prof. Larry Pileggi. Her research is a novel combination of optimization and machine learning tools to innovate the system modeling and estimation for improving resiliency of the next generation cyber-physical Infrastructure against natural and adversarial threats. Her work has been part of SUGAR toolbox which is a power grid analysis tool commercialized by Pearl Street Technologies, Inc. Her work has been recognized by the best paper award in the 2021 PES general meeting, and selected by 2023 Microsoft Accelerate Foundation Models Research Initiative.