Feasibility of Multiplex Communication in a 2D Mesh Asynchronous Neural Network with Fluctuations

Abstract

It remains a mystery how neural networks composed of neurons with fluctuating characteristics can reliably transmit information. In this study, we simulated a 9 × 9 2D mesh neural network consisting of an integrate-and-fire model without leak, and connection weights that were randomly generated. The characteristics of the refractory period and output delay of the neurons were fluctuated time to time. Spikes from transmitting neuron groups spread (propagated as spike waves) to receiving neurons. For 9 to 1 multiplex communication with a back propagation neural network (BPN), the receiving neurons successfully classified which neuron group transmitted the spike at a rate of 99%. In other words, the activity of the neuron group is propagated in the neural network as spike waves in a broadcasting manner and the wave fragment is received by receiving neurons. Next, point-to-point signal transmission in the neural network is carried out by multi-path, multiplex communication, and diversity reception. Each neuron can function in 3 ways of transmit, relay (transfer), and receive; however, most neurons act as a local relaying media. This type of mechanism is similar to sound propagation through air. Our research group studies the functions of neural networks by combining experiments with cultured neuronal networks with artificial neural network simulations. This current study corresponding to our previous work on the ability of remote receiving neurons to identify two transmitting neuron groups stimulated in a cultured neuronal network, i.e., 2 to 1 communication. These mechanisms may be the basis of higher cortical functions.

Keywords

• neural network; multiplex communication; BPN; learning; fluctuation of neuron