Title: Towards Private Information Design at Scale
Date and time: 12 June 2025 (Thursday), 3:30 – 4:30 pm
Venue: IEOR Seminar Hall
Speaker: Ketan Savla, University of Southern California
Abstract: Congestion pricing has been a key tool to spread traffic demand across space and time. Effective pricing however requires knowing personal attributes, such as values of time. We present a “revealed preference” based algorithm to learn these attributes from historical routing decisions and demonstrate its advantage over classical “stated preference” method in human subject experiments.
Non-monetary mechanisms such as travel recommendations currently are strongly correlated among travelers and face the prospect of inverse U-shaped social welfare. Bayesian information design is a compelling framework for generating persuasive private recommendations. Feasibility of a recommendation policy is often characterized by the “obedience constraint” according to which following recommendation is, in a posteriori expectation, no worse than any other action. The probabilistic recommendation policy has continuous support for non-atomic games. This makes it computationally challenging to handle the obedience constraint for optimal design, and further questions the practical validity of Bayesian computation in decision-making for non-atomic setting. We address these issues in the context of routing games. We also study a repeated setting, in which the likelihood of an agent following a recommendation in a stage is proportional to the population-wide average regret from previous stages, e.g., provided by a review aggregator platform. We present results from a human subject experiment on the validity of this “learning model” and establish its convergence to Bayes correlated equilibrium under a fixed obedient recommendation policy.
Bio: Ketan Savla is an associate professor and the John and Dorothy Shea Early Career Chair in Civil Engineering at the University of Southern California. His current research interest is in distributed optimal and robust control, dynamical networks, state-dependent queuing systems, and mechanism design, with applications in civil infrastructure and autonomous systems. His recognitions include NSF CAREER, IEEE CSS George S. Axelby Outstanding Paper Award, AACC Donald P. Eckman Award, and the IEEE ITS Outstanding Application Award. He serve(d) as an associate editor of the IEEE Transactions on Control of Network Systems, IEEE Control Systems Letters (L-CSS), and IEEE Transactions on Intelligent Transportation Systems. He currently serves as a senior editor of the L-CSS. He is also a co-founder and the chief science officer of Xtelligent, Inc.