Bayesian MLP neural networks for image analysis

We demonstrate the advantages of using Bayesian multi-layer perceptron (MLP ) neural networks for image analysis. The Bayesian approach provides consis tent way to do inference by combining the evidence from the data to prior k nowledge from the problem. A practical problem with MLPs is to select the c orrect complexity for the model, i.e., the right number of hidden units or correct regularization parameters. The Bayesian approach offers efficient t ools for avoiding overfitting even with very complex models, and facilitate s estimation of the confidence intervals of the results. In this contributi on we review the Bayesian methods for MLPs and present comparison results f rom two case studies. In the first case, MLPs were used to solve the invers e problem in electrical impedance tomography. The Bayesian MLP provided con sistently better results than other methods. In the second case, the goal w as to locate trunks of trees in forest scenes. With Bayesian MLP it was pos sible to use large number of potentially useful features and prior for dete rmining the relevance of the features automatically. (C) 2000 Elsevier Scie nce B.V. All rights reserved.