A general neural membrane computing model

Abstract

Neural P systems are a class of distributed parallel computing models inspired by the way neurons cooperate in biological neural networks. Artificial neural networks integrate biological neural networks with mathematical models, which have powerful information processing capabilities. This work adopts the structure and data processing method of neural networks and integrates them with membrane computing to improve the neural P systems in data processing. A general neural membrane computing (GNMC) model is developed to realize the flow and manipulation of data in the form of objects and rules in the neuron. Glial cells are introduced into the GNMC model at the cellular level to modulate the states of the connected neurons. The trigger conditions, the activation mechanisms and the dissolve-reconnect rules make the GNMC model more efficient in handling objects. The multiple objects and multiple topological structures enable the GNMC model to handle different practical problems. The Turing universality of the GNMC model is proved as number generating and accepting devices. Furthermore, a small universal GNMC model with 71 neurons, fewer than that of the existing state-of-the-art neural P systems, is constructed for computing functions.