Empirical Mechanism Design: Designing Mechanisms from Data

Enrique Areyan Viqueira, Cyrus Cousins, Yasser Mohammad, Amy Greenwald
Proceedings of The 35th Uncertainty in Artificial Intelligence Conference, PMLR 115:1094-1104, 2020.

Abstract

We introduce a methodology for the design of parametric mechanisms, which are multiagent systems inhabited by strategic agents, with knobs that can be adjusted to achieve specific goals. We assume agents play approximate equilibria, which we estimate using the probably approximately correct learning framework. Under this assumption, we further learn approximately optimal mechanism parameters. We do this theoretically, assuming a finite design space, and heuristically, using Bayesian optimization (BO). Our BO algorithm incorporates the noise associated with modern concentration inequalities, such as Hoeffding’s, into the underlying Gaussian process. We show experimentally that our search techniques outperform standard baselines in a stylized but rich model of advertisement exchanges.

Cite this Paper

BibTeX
@InProceedings{pmlr-v115-viqueira20a, title = {Empirical Mechanism Design: Designing Mechanisms from Data}, author = {Viqueira, Enrique Areyan and Cousins, Cyrus and Mohammad, Yasser and Greenwald, Amy}, booktitle = {Proceedings of The 35th Uncertainty in Artificial Intelligence Conference}, pages = {1094--1104}, year = {2020}, editor = {Adams, Ryan P. and Gogate, Vibhav}, volume = {115}, series = {Proceedings of Machine Learning Research}, month = {22--25 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v115/viqueira20a/viqueira20a.pdf}, url = {https://proceedings.mlr.press/v115/viqueira20a.html}, abstract = {We introduce a methodology for the design of parametric mechanisms, which are multiagent systems inhabited by strategic agents, with knobs that can be adjusted to achieve specific goals. We assume agents play approximate equilibria, which we estimate using the probably approximately correct learning framework. Under this assumption, we further learn approximately optimal mechanism parameters. We do this theoretically, assuming a finite design space, and heuristically, using Bayesian optimization (BO). Our BO algorithm incorporates the noise associated with modern concentration inequalities, such as Hoeffding’s, into the underlying Gaussian process. We show experimentally that our search techniques outperform standard baselines in a stylized but rich model of advertisement exchanges.} }
Endnote
%0 Conference Paper %T Empirical Mechanism Design: Designing Mechanisms from Data %A Enrique Areyan Viqueira %A Cyrus Cousins %A Yasser Mohammad %A Amy Greenwald %B Proceedings of The 35th Uncertainty in Artificial Intelligence Conference %C Proceedings of Machine Learning Research %D 2020 %E Ryan P. Adams %E Vibhav Gogate %F pmlr-v115-viqueira20a %I PMLR %P 1094--1104 %U https://proceedings.mlr.press/v115/viqueira20a.html %V 115 %X We introduce a methodology for the design of parametric mechanisms, which are multiagent systems inhabited by strategic agents, with knobs that can be adjusted to achieve specific goals. We assume agents play approximate equilibria, which we estimate using the probably approximately correct learning framework. Under this assumption, we further learn approximately optimal mechanism parameters. We do this theoretically, assuming a finite design space, and heuristically, using Bayesian optimization (BO). Our BO algorithm incorporates the noise associated with modern concentration inequalities, such as Hoeffding’s, into the underlying Gaussian process. We show experimentally that our search techniques outperform standard baselines in a stylized but rich model of advertisement exchanges.
APA
Viqueira, E.A., Cousins, C., Mohammad, Y. & Greenwald, A.. (2020). Empirical Mechanism Design: Designing Mechanisms from Data. Proceedings of The 35th Uncertainty in Artificial Intelligence Conference, in Proceedings of Machine Learning Research 115:1094-1104 Available from https://proceedings.mlr.press/v115/viqueira20a.html.

Related Material