Computing Topological Indices of 3-Layered Artificial Neural Network

Abstract

Let η be a network graph with vertex and edge sets P(η) and E(η), respectively. This study aims to find the expected value (obtained during training and testing data) for Artificial Neural Networks (ANN) through indices. A three-layer artificial neural network is considered here, which we call ANN(m, n, o). Moreover, a comparison is given between the topological indices (TI) of ANN with topological indices (TI) of the Probabilistic Neural Network (PNN). By comparing the indices, we can assess the effect of network structure on ANN model accuracy. The comparison between the two approaches helps us understand the accuracy and performance of ANN and PNN models. We can also gain insights into the differences between ANN and PNN in terms of their ability to learn and generalize.