A Prediction Model for Network Information Dissemination based on Optimized Convolutional Neural Network Using Enhanced Kepler Optimization

The information dissemination methodology effectively measures the spread of opinions in Social Networks (SNs). Because of the unpredictability, irregularity of data, complexity, typical data distribution index systems rely heavily on human intervention, have substantial deviations, and are used in a narrow range. In prior researches, a Convolutional Neural Network (CNN)-based strategy is created from critical viewpoints for creating a multilayer and multigranular information dissemination index system. However, as information grows, the CNN requires the best parameter specifications, as well as a more specialized system for particular information dissemination events on various themes. Thus, an Enhanced Kepler Optimization Algorithm (EKOA) approach is developed to tune the hyperparameters of the CNN for the prediction of network information dissemination. Then the weight is trained, and it is used to determine the weight of information dissemination methods. The simulation findings show that EKOA-CNN achieves 99.98% and outperforms other approaches such as game inference cellular automaton model (SEInR), and CNN.