Learning Quadratic Games on Networks

Yan Leng, Xiaowen Dong, Junfeng Wu, Alex Pentland
Proceedings of the 37th International Conference on Machine Learning, PMLR 119:5820-5830, 2020.

Abstract

Individuals, or organizations, cooperate with or compete against one another in a wide range of practical situations. Such strategic interactions are often modeled as games played on networks, where an individual’s payoff depends not only on her action but also on that of her neighbors. The current literature has largely focused on analyzing the characteristics of network games in the scenario where the structure of the network, which is represented by a graph, is known beforehand. It is often the case, however, that the actions of the players are readily observable while the underlying interaction network remains hidden. In this paper, we propose two novel frameworks for learning, from the observations on individual actions, network games with linear-quadratic payoffs, and in particular, the structure of the interaction network. Our frameworks are based on the Nash equilibrium of such games and involve solving a joint optimization problem for the graph structure and the individual marginal benefits. Both synthetic and real-world experiments demonstrate the effectiveness of the proposed frameworks, which have theoretical as well as practical implications for understanding strategic interactions in a network environment.

Cite this Paper

BibTeX
@InProceedings{pmlr-v119-leng20a, title = {Learning Quadratic Games on Networks}, author = {Leng, Yan and Dong, Xiaowen and Wu, Junfeng and Pentland, Alex}, booktitle = {Proceedings of the 37th International Conference on Machine Learning}, pages = {5820--5830}, year = {2020}, editor = {III, Hal Daumé and Singh, Aarti}, volume = {119}, series = {Proceedings of Machine Learning Research}, month = {13--18 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v119/leng20a/leng20a.pdf}, url = {https://proceedings.mlr.press/v119/leng20a.html}, abstract = {Individuals, or organizations, cooperate with or compete against one another in a wide range of practical situations. Such strategic interactions are often modeled as games played on networks, where an individual’s payoff depends not only on her action but also on that of her neighbors. The current literature has largely focused on analyzing the characteristics of network games in the scenario where the structure of the network, which is represented by a graph, is known beforehand. It is often the case, however, that the actions of the players are readily observable while the underlying interaction network remains hidden. In this paper, we propose two novel frameworks for learning, from the observations on individual actions, network games with linear-quadratic payoffs, and in particular, the structure of the interaction network. Our frameworks are based on the Nash equilibrium of such games and involve solving a joint optimization problem for the graph structure and the individual marginal benefits. Both synthetic and real-world experiments demonstrate the effectiveness of the proposed frameworks, which have theoretical as well as practical implications for understanding strategic interactions in a network environment.} }
Endnote
%0 Conference Paper %T Learning Quadratic Games on Networks %A Yan Leng %A Xiaowen Dong %A Junfeng Wu %A Alex Pentland %B Proceedings of the 37th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2020 %E Hal Daumé III %E Aarti Singh %F pmlr-v119-leng20a %I PMLR %P 5820--5830 %U https://proceedings.mlr.press/v119/leng20a.html %V 119 %X Individuals, or organizations, cooperate with or compete against one another in a wide range of practical situations. Such strategic interactions are often modeled as games played on networks, where an individual’s payoff depends not only on her action but also on that of her neighbors. The current literature has largely focused on analyzing the characteristics of network games in the scenario where the structure of the network, which is represented by a graph, is known beforehand. It is often the case, however, that the actions of the players are readily observable while the underlying interaction network remains hidden. In this paper, we propose two novel frameworks for learning, from the observations on individual actions, network games with linear-quadratic payoffs, and in particular, the structure of the interaction network. Our frameworks are based on the Nash equilibrium of such games and involve solving a joint optimization problem for the graph structure and the individual marginal benefits. Both synthetic and real-world experiments demonstrate the effectiveness of the proposed frameworks, which have theoretical as well as practical implications for understanding strategic interactions in a network environment.
APA
Leng, Y., Dong, X., Wu, J. & Pentland, A.. (2020). Learning Quadratic Games on Networks. Proceedings of the 37th International Conference on Machine Learning, in Proceedings of Machine Learning Research 119:5820-5830 Available from https://proceedings.mlr.press/v119/leng20a.html.

Related Material