Instabilities in learning automata playing games with delayed information

Author

Billard, Edward A.

Author_Institution

Fac. of Comput. Sci. & Eng., Aizu Univ., Japan

Volume

2

fYear

1994

fDate

2-5 Oct 1994

Firstpage

1160

Abstract

Previous results in learning automata playing sequential stochastic games showed that, with the proper choice of parameters, the players learn the optimal mixed strategy. The model is extended to games with delayed information concerning the other player´s mixed strategy, as might result from the latencies in a distributed system. Predicted oscillations from a nonlinear delay differential equation and from simulations are presented. An outline of a linear stability analysis shows that as parameters are chosen to more closely approximate the optimal strategies, the system is more susceptible to delay-initiated instabilities

Keywords

delays; learning automata; nonlinear differential equations; oscillations; stability; stochastic games; delayed information; learning automata instabilities; linear stability analysis; nonlinear delay differential equation; optimal mixed strategy; optimal strategy approximation; predicted oscillations; sequential stochastic games; Computer science; Delay; Delay effects; Differential equations; Distributed decision making; Game theory; Learning automata; Predictive models; Stability analysis; Stochastic processes; Uncertainty;

fLanguage

English

Publisher

ieee

Conference_Titel

Systems, Man, and Cybernetics, 1994. Humans, Information and Technology., 1994 IEEE International Conference on

Conference_Location

San Antonio, TX

Print_ISBN

0-7803-2129-4

Type

conf

DOI

10.1109/ICSMC.1994.400001

Filename

400001