Abstract: Cyber-physical systems (CPS) require the construction and management of various models to assure their correct, safe, and secure operation. The seminar will first explore CPS applications and the various formalisms present in current CPS research. Unfortunately, to date the different model views of cyber-physical systems are disparate and largely related informally, which raises issues with the degree of formal consistency between those various models of requirements, system behavior, and system architecture. The talk will then present a category-theoretic framework that makes different types of composition explicit in the modeling and analysis of cyber-physical systems, which assist in verifying the system as a whole. In particular, the seminar will focus on modeling system behavior via algebras on the monoidal category of wiring diagrams, system requirements (e.g. on safety) again as an algebra on certain objects in this monoidal category, and the CPS design process using the notion of a slice category.
Finally, the talk will pivot and present a series of recent results in machine learning involving the use of decision trees and their surprising competitiveness with deep neural networks across a series of tasks including uncertainty quantification, reinforcement learning, and safe control.
Bio: Cody Fleming is an associate professor in Mechanical Engineering at Iowa State University. Dr. Fleming is interested in developing theory and methods to assure the safety and security of controls systems and autonomy more broadly. His application areas include space systems, air traffic management, urban air mobility, and autonomous driving. He received his PhD in Aeronautics and Astronautics from MIT and before joining Iowa State, he was on the faculty at the University of Virginia.