Feed-forward neural networks (Multi-Layered Perceptrons) are used widely in real-world regression or classification tasks. A reliable and practical measure of prediction "confidence" is essential in real-world tasks. This paper compares three approachesto prediction confidenceestimation, using both artificial and real data. The three methods are maximum likelihood, approximate Bayesian and bootstrap. Both noiseinherent to the dataand model uncertaintyare considered.
Georgios Papadopoulos, Peter J. Edwards, Alan F. M