стохастична гра

The problem of dynamic coordination of strategies of multiagent systems in the conditions of uncertainty on the basis of stochastic game model is solved. Dynamic coordination consists in system learning to generate the spatially-distributed periodic signals. The model of stochastic game is constructed, criteria of dynamic coordination of strategies are defined, a recurrent method, algorithmic and software support for the stochastic game solving are developed.

The problem of reinforcement learning of multiagent systems in the game formulation is considered. The Markovian model of stochastic game is constructed, criteria of game learning are formulated, the Q-method and corresponding algorithm of the stochastic game solving are described, results of computer realization of a Q-method are analyzed.

The adaptive game method of events synchronization in multiagent systems in the conditions of uncertainty is developed. The essence of a method consists in alignment of delays of approach of events on the basis of supervision of actions of the next players. The formulation of stochastic game is executed and game algorithm for its solving is developed. Influences of parameters on convergence of a game method are investigated by means of computer experiment.

Розроблена модель матричної стохастичної гри для прийняття рішень в умовах невизначеності. Запропоновано метод Q-навчання для розв’язування стохастичної гри з апріорі невідомими матрицями виграшів. Виконано формулювання ігрової задачі, описано марківський рекурентний метод та алгоритм для її розв’язування. Отримано та проаналізовано результати комп’ютерного моделювання стохастичної гри з Q-навчанням.