Invited Talk
The Interplay of Machine Learning and Mechanism Design
David Parkes

Wed Dec 8th 08:30 -- 09:20 AM @ Regency Ballroom

In the economic theory of mechanism design, the goal is to elicit private information from each of multiple agents in order to select a desirable system wide outcome, and despite agent self-interest in promoting individually beneficial outcomes. Auctions provide a canonical example, with information elicited in the form of bids, and an allocation of resources and payments defining an outcome. Indeed, one aspect of the emerging interplay between machine learning (ML) and mechanism design (MD) arises by interpreting auctions as a method for learning agent valuation functions. In addition to seeking sufficient accuracy to support optimal resource allocation, we require for incentive compatibility that prices are insensitive to the inputs of any individual agent and find an interesting connection with regularization in statistical ML. More broadly, ML can be used for de novo design, in learning payment rules with suitable incentive properties. Ideas from MD are also flowing into ML. One example considers the use of mechanisms to elicit private state, reward and transition models, in enabling coordinated exploration and exploitation in multi-agent systems despite self-interest. Another application is to supervised learning, where labeled training data is elicited from self-interested agents, each with its own objective criterion on the hypothesis learned by the mechanism. Looking ahead, a tantalizing challenge problem is to adopt incentive mechanisms for the design of robust agent architectures, for example in assigning internal rewards that promote modular intelligent systems.

Author Information

David Parkes (Harvard University)

David C. Parkes is Gordon McKay Professor of Computer Science in the School of Engineering and Applied Sciences at Harvard University. He was the recipient of the NSF Career Award, the Alfred P. Sloan Fellowship, the Thouron Scholarship and the Harvard University Roslyn Abramson Award for Teaching. Parkes received his Ph.D. degree in Computer and Information Science from the University of Pennsylvania in 2001, and an M.Eng. (First class) in Engineering and Computing Science from Oxford University in 1995. At Harvard, Parkes leads the EconCS group and teaches classes in artificial intelligence, optimization, and topics at the intersection between computer science and economics. Parkes has served as Program Chair of ACM EC’07 and AAMAS’08 and General Chair of ACM EC’10, served on the editorial board of Journal of Artificial Intelligence Research, and currently serves as Editor of Games and Economic Behavior and on the boards of Journal of Autonomous Agents and Multi-agent Systems and INFORMS Journal of Computing. His research interests include computational mechanism design, electronic commerce, stochastic optimization, preference elicitation, market design, bounded rationality, computational social choice, networks and incentives, multi-agent systems, crowd-sourcing and social computing.

More from the Same Authors