On Eco-Grammar Systems and Artificial Neural Networks

Eco-grammar systems and artificial neural networks have many common features: massive parallelism, independently working elements (agents/neurons), cooperation of the elements and, not least, universal computational power (at least as that of Turing Machine). We prove the possibility to simulate each step of the one type system by the fixed number of steps of the another type one without loss of the parallelism. Moreover, number of processing elements (neurons, agents) of the model is a function of class O(n), where n is a number of processing elements of the original system. Keywords: grammar systems, EG systems, artificial neural networks, Turing machine, computability. 1 Introduction The theoretical possibility of simulating of artificial neural networks (ANN) models [18],[5] by eco-grammar (EG) systems [6], [3] (and conversely) have been known for few years, due to proofs of their universal computational power [3], [16]. However, the principles of these proofs do not allow direct..