comparison toolboxes/FullBNT-1.0.7/netlab3.3/rbfprior.m @ 0:e9a9cd732c1e tip

first hg version after svn

+function [mask, prior] = rbfprior(rbfunc, nin, nhidden, nout, aw2, ab2)
+%RBFPRIOR Create Gaussian prior and output layer mask for RBF.
+%
+%	Description
+%	[MASK, PRIOR] = RBFPRIOR(RBFUNC, NIN, NHIDDEN, NOUT, AW2, AB2)
+%	generates a vector MASK  that selects only the output layer weights.
+%	This is because most uses of RBF networks in a Bayesian context have
+%	fixed basis functions with the output layer as the only adjustable
+%	parameters.  In particular, the Neuroscale output error function is
+%	designed to work only with this mask.
+%
+%	The return value PRIOR is a data structure,  with fields PRIOR.ALPHA
+%	and PRIOR.INDEX, which specifies a Gaussian prior distribution for
+%	the network weights in an RBF network. The parameters AW2 and AB2 are
+%	all scalars and represent the regularization coefficients for two
+%	groups of parameters in the network corresponding to  second-layer
+%	weights, and second-layer biases respectively. Then PRIOR.ALPHA
+%	represents a column vector of length 2 containing the parameters, and
+%	PRIOR.INDEX is a matrix specifying which weights belong in each
+%	group. Each column has one element for each weight in the matrix,
+%	using the standard ordering as defined in RBFPAK, and each element is
+%	1 or 0 according to whether the weight is a member of the
+%	corresponding group or not.
+%
+%	See also
+%	RBF, RBFERR, RBFGRAD, EVIDENCE
+%
+
+%	Copyright (c) Ian T Nabney (1996-2001)
+
+nwts_layer2 = nout + (nhidden *nout);
+switch rbfunc
+case 'gaussian'
+   nwts_layer1 = nin*nhidden + nhidden;
+case {'tps', 'r4logr'}
+   nwts_layer1 = nin*nhidden;
+otherwise
+   error('Undefined activation function');
+end  
+nwts = nwts_layer1 + nwts_layer2;
+
+% Make a mask only for output layer
+mask = [zeros(nwts_layer1, 1); ones(nwts_layer2, 1)];
+
+if nargout > 1
+  % Construct prior
+  indx = zeros(nwts, 2);
+  mark2 = nwts_layer1 + (nhidden * nout);
+  indx(nwts_layer1 + 1:mark2, 1) = ones(nhidden * nout, 1);
+  indx(mark2 + 1:nwts, 2) = ones(nout, 1);
+
+  prior.index = indx;
+  prior.alpha = [aw2, ab2]';
+end