A Predictive Model for Uncertainty Analysis on Big Data Using Bayesian Convolutional Neural Network (CNN)

Abstract

Abstract

The need of addressing uncertainty in big data increases as more data is created and examined. It is essential to comprehend, measure, and control uncertainty in large data for dependable and useful analysis. Uncertainty in big data analysis is one of the major problems of big data, and if not handled correctly, it will lead to wrong predictions/classification of the model. In order to solve the problem of uncertainty in big data, this paper presents a Bayesian CNN model for the prediction of uncertainty in big data. The Bayesian CNN model uses a probability score in predicting uncertainties in big data. With this, it does not just show the classified results that were made by the model, it also shows the probability score, which signifies the decision score of the model when making classifications on images. The result of Bayesian model shows a better result of 99.9% for both training and testing.