Improvement of Multi-Layer Perceptron by Pulse Glial Network with Dynamic Period of Inactivity

In this study, we propose a Multi-Layer Perceptron (MLP) with pulse glial network having dynamic period of inactivity. We connect glias to neurons in a hidden-layer. The glia is excited by the connecting neuron output. Then, the glia generates a pulse. The pulse is propagated to the connecting neuron and the neighboring glia. In the previous method, we fix a period of inactivity. The period of inactivity decides the frequency of the pulse generation. Thereby, the pulse generation pattern often becomes periodic. We consider that it is similar to the local minimum. In this network, the period of inactivity of the glia is dynamically changed according to a pulse generation time. By varying the period of inactivity, the pulse generation pattern obtains the diversity. We consider that this diversity of the pulse generation pattern is efficiency to the MLP performance. By the simulation, we confirm that the proposed MLP improves the MLP performance than the conventional MLP.