Game Theory and Learning for Cyber-Physical Systems


  • Tansu Alpcan
  • Tamer Başar
  • José M. Maestre
  • Quanyan Zhu


  • Tansu Alpcan, University of Melbourne, Australia
  • Tamer Başar, University of Illinois, USA
  • José María Maestre, University of Seville, Spain
  • Quanyan Zhu, New York University, USA


Machine and Deep Learning have enjoyed increasing popularity in recent years by successfully addressing a variety of long-standing problems in Artificial Intelligence. However, their applications to Cyber-Physical Systems (CPSs) and Internet-of-Things (IoT) have been limited until now. Game Theory has been a popular approach to address a range of engineering problems, especially in distributed systems, control, and communications. Its computational scalability, which is crucial for modern CPS and IoT applications, remains mostly as an open research topic. The synthesis of computational methods from machine and deep learning with game-theoretic methods for distributed decision making provides a powerful combination when applied to 21st century systems under the umbrella of CPS and IoT. Examples come from various different fields, from medical devices and robots on a small scale, to power systems and connected communities on a large scale. Since CPSs are expected to operate in dynamically changing environments, new analytical and design tools using machine learning and game theory would be needed to significantly improve their operational efficiency, harden the security, enhance resiliency, and address human and socio-economic aspects of CPSs.


  • Introduction to learning and games for cyber-physical systems, Tamer Başar
  • Deep learning and security games for cyber-physical systems, Tansu Alpcan
  • High-confidence learning for secure control of cyber-physical systems, Quanyan Zhu
  • Data-driven distributed model predictive control, José María Maestre