RBF networks for source localization in quantitative electrophysiology

The backpropagation neural network methods have been proposed recently to s olve the inverse problem in quantitative electrophysiology.(2) A major adva ntage of the technique is that once a neural network is trained, it no long er requires iterations or access to sophisticated computations. We propose to use RBF networks for source localization in the brain, and systematicall y compare their performance to those of Levenberg-Marquardt (LM) algorithms . We show the use of two types of Radial Basis Function Networks (RBF) netw ork: a classic network with fixed number of hidden layer neurons(4) and an improved network, Minimal Resource Allocation Network (MRAN),(5) recently p roposed by one of the authors, capable for dynamically configuring its stru cture so as to obtain a compact topology to match the data presented to it.