Mon, Jul 31, 2017 @ 02:00 PM - 03:00 PM
Ming Hsieh Department of Electrical and Computer Engineering
Conferences, Lectures, & Seminars
Speaker: Dario Paccagnan, ETH Zurich
Talk Title: Distributed optimization through game design
Abstract: A fundamental challenge in multiagent systems is to design local control algorithms to ensure a desirable collective behaviour. In recent years, game theory has emerged as a valuable language to decouple a system-level objective into local utility functions to assign to each agent. In this talk, we consider a class of resource allocation problems and exemplify how tools from game theory can be leveraged to produce distributed algorithms with guaranteed performance, even in the presence of uncertainty. Our main contribution consists in showing how an improvement in the worst case performance comes at the expenses of the best case one. We conclude by discussing how this fundamental tradeoff can be overcome if we allow local algorithms to depend on a higher degree of system-level information.
Biography: Dario Paccagnan is a doctoral student at the Automatic Control Laboratory, ETH Zurich since January 2015 under the supervision of Prof. John Lygeros. He is currently visiting the University of California, Santa Barbara hosted by Prof. Jason Marden. He earned his Laurea (B.Sc.) and Laurea Magistrale (M.Sc.) in Aerospace Engineering from the University of Padova in 2011 and 2014. In the same year he received the M.Sc. in Mathematical Modelling from the Technical University of Denmark, all with Honours. From January to August 2014 he has been a visiting scholar at Imperial College of London, hosted by Prof. Alessandro Astolfi. Dario's research interests lie at the intersection between Game Theory and Distributed Control. Applications include resource allocation problems, optimisation of energy systems and control of traffic networks.
Host: Prof. Mihailo Jovanovic
Audiences: Everyone Is Invited
Contact: Annie Yu