Effect of Degree Distribution on Signal Propagation Based on Heterogeneous High-frequency Stimuli in Feedforward Networks

Abstract

In the feedforward neural network based on FHN neurons, heterogeneous high-frequency signals have a promoting effect on weak low-frequency signals, and the influence of connection probability and connection strength on neuron firing rate in the feedforward neural network with three different degree distributions (identical, uniform and exponential) under the action of heterogeneous high-frequency stimuli is studied in this paper. It is found that when the connection probability is low, the weak frequency signal in the topology with the identical distribution is more affected by the high-frequency stimulus than the other two distributions, but when the amplitude of the high-frequency signal is too low, the high-frequency signal has no effect on the weak low-frequency signal, and even the firing rate is zero. However, the high-frequency stimulus with low amplitude has a promoting effect on the weak signal, and the propagation of the weak signal in the three distributions increases with the enhancement of the high-frequency stimulus amplitude. As the amplitude of high-frequency increases gradually, the promotion effect of high-frequency on weak signals under uniform and exponential distribution is gradually balanced. On the contrary, the promotion effect of high-frequency on weak signals under the identical distribution always enhances with the increases of connection probability. Meanwhile, the different connection strength of the three networks will also have different effects on the signal firing. This result shows that the influence of high-frequency stimulus on weak signal is not only affected by frequency amplitude, but also related to network topology, connection probability and connection strength.