Continuous optimization using a hybrid model of cellular automata and learning automata

Authors: Evsyutin O. O., Shelupanov A. A., Babishin V. D., Cosedko K. A.

Annotation: The article presents an algorithm for continuous optimization of several variables functions, based on a computational model of cellular automata with an objective function. An effectiveness of this computational model depends on choice of the cellular automata development rule. At the same time, the use of rule compositions shows greater efficiency compared with the use of individual rules. A distinctive feature of this study is the use of a dynamic composition, formed during the development of cellular automata with an objective function. The rule choice at each step of the cellular automata development is carried out by a learning automata computational model. The results of computational experiments conducted with standard test functions show that this solution can improve the accuracy of optimization.

Keywords: continuous optimization, cellular automata, learning automata, cellular automaton with an objective function