CAREER: Decision Making, Learning, and Incentive Design in Multilayer Networks

点击翻译按钮获取中文摘要

Networks are prevalent in all aspects of society. They play a central role in information transmission, social interactions, economic transactions, and the spread of diseases, to name a few. Moreover, many networks, including this underlying critical infrastructure such as communication, power, and transportation networks, are highly interdependent. While there exists considerable research on the modeling and analysis of isolated networks, the study of interdependent and multilayer networks, particularly in the presence of strategic and self-interested decision makers, confronts many open challenges. This proposal will leverage tools from optimization, game theory, machine learning, and graph theory towards building an analytical framework for the study of multilayer networks. The findings will have impact on the design and operation of self-organizing multi-agent systems in complex network environments, as well as in securing interdependent critical infrastructure. Integrated with the proposed research, the education plan will broaden participation in the proposed research areas and provide audiences at all levels with applied and hands-on experiences through collaboration with outreach programs at the Ohio State University. At the core of this proposal is the study of multilayer networks using game-theoretic modeling and analysis. This project will propose and analyze a new class of multi-network games for the study of decentralized decision-making and learning over interconnected networks. This framework will account for the multi-modality of information and communication channels available to entities in multi-agent systems and identify the potential sources of inefficiency in decentralized decision making in multilayer network environments (as compared to both centrally operated and single layer networks). It will also propose and evaluate the impact of incentives or interventions to shape the outcomes of decentralized decision making and distributed learning on these networks, through mechanism design and targeted interventions. These findings will be used to show how the holistic study of multilayer networks, as opposed to focusing on each network individually, enhances our ability to design and evaluate economic and regulatory interventions, and prevent unwanted equilibria or learning outcomes.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.