Some Comments and Experimental Results on Bayesian Regularization
This paper attempts to determine whether the Bayesian paradigm could improve Multi-Layer Perceptrons (MLPs) learning procedures. In particular, it deals with the paradigm developed by D MacKay. Critical points of MacKay's work are highlighted. Bayesian MLPs are used on three public classification databases and compared to other methods.
Bibliographic Reference: Paper presented: Eighth Italian Workshop on Neural Networks, Vietri Sul Mare (IT), May 22-24, 1996
Availability: Available from (1) as Paper EN 39752 ORA
Record Number: 199610669 / Last updated on: 1996-06-21
Original language: en
Available languages: en