When reasoning in the presence of uncertainty there is a unique and self-consistent set of rules for induction and model selection – Bayesian inference. Recent advances in neural networks have been fuelled by the adoption of this Bayesian framework, either implicitly, for example through the use of committees, or explicitly through Bayesian evidence and sampling frameworks. In this chapter, we show how this ‘second generation’ of neural network techniques can be applied to biomedical data and focus on the networks’ ability to provide assessments of the confidence associated with its predictions. This is an essential requirement for any automatic biomedical pattern recognition system. It allows low confidence decisions to be highlighted and deferred, possibly to a human expert, and falls naturally out of the Bayesian framework.
[1]
Geoffrey E. Hinton,et al.
Bayesian Learning for Neural Networks
,
1995
.
[2]
Christopher M. Bishop,et al.
Neural networks for pattern recognition
,
1995
.
[3]
Brian D. Ripley,et al.
Neural Networks and Related Methods for Classification
,
1994
.
[4]
J. M. Spyers-Ashby,et al.
A comparison of fast fourier transform (FFT) and autoregressive (AR) spectral estimation techniques for the analysis of tremor data
,
1998,
Journal of Neuroscience Methods.
[5]
David J. C. MacKay,et al.
The Evidence Framework Applied to Classification Networks
,
1992,
Neural Computation.
[6]
Stephen J. Roberts,et al.
Maximum certainty approach to feedforward neural networks
,
1997
.