Evolutionary-estimated programming the Turing machine by differential evolution

Abstract

This research deals with employing an artificial intelligence at estimating the rules for programming the Turing machine. The aim of this project is to simplify a programming process, which consists of designing the Turing machine's transition function. This is realized by using artificial intelligence for performing composition of the rules which map responses of the transition function in dependence on its arguments. Within the research, there were chosen a few of suitable artificial intelligence algorithms for solving this problem. In the previous working on this project, Self-Organizing Migrating Algorithm had been used. As another, there was chosen Differential Evolution as the suitable artificial intelligence algorithm for solving this problem. In this paper there is described an employment of the latter algorithm at estimating the rules for programming the Turing machines. This paper subsequently proves that the evolution can produce valid rules for solving concrete tasks by the Turing machine. This research has confirmed possibilities of simplifying programming the Turing machine by artificial intelligence.