Bayesian neural networks for industrial applications

Abstract

Demonstrates the advantages of using Bayesian neural networks in regression, inverse and classification problems, which are common in industrial applications. The Bayesian approach provides a consistent way to perform inference by combining the evidence from data with prior knowledge from the problem. A practical problem with neural networks is to select the correct complexity for the model, i.e. the right number of hidden units or correct regularization parameters. The Bayesian approach offers efficient tools for avoiding overfitting, even with very complex models, and facilitates the estimation of the confidence intervals of the results. In this paper, we review the Bayesian methods for neural networks and present comparison results from case studies in the prediction of the quality properties of concrete (regression), electrical impedance tomography (inverse problem) and forest scene analysis (classification). The Bayesian networks provided consistently better results than other methods.