annotate toolboxes/FullBNT-1.0.7/netlab3.3/mlpunpak.m @ 0:e9a9cd732c1e tip

first hg version after svn
author wolffd
date Tue, 10 Feb 2015 15:05:51 +0000
parents
children
rev   line source
wolffd@0 1 function net = mlpunpak(net, w)
wolffd@0 2 %MLPUNPAK Separates weights vector into weight and bias matrices.
wolffd@0 3 %
wolffd@0 4 % Description
wolffd@0 5 % NET = MLPUNPAK(NET, W) takes an mlp network data structure NET and a
wolffd@0 6 % weight vector W, and returns a network data structure identical to
wolffd@0 7 % the input network, except that the first-layer weight matrix W1, the
wolffd@0 8 % first-layer bias vector B1, the second-layer weight matrix W2 and the
wolffd@0 9 % second-layer bias vector B2 have all been set to the corresponding
wolffd@0 10 % elements of W.
wolffd@0 11 %
wolffd@0 12 % See also
wolffd@0 13 % MLP, MLPPAK, MLPFWD, MLPERR, MLPBKP, MLPGRAD
wolffd@0 14 %
wolffd@0 15
wolffd@0 16 % Copyright (c) Ian T Nabney (1996-2001)
wolffd@0 17
wolffd@0 18 % Check arguments for consistency
wolffd@0 19 errstring = consist(net, 'mlp');
wolffd@0 20 if ~isempty(errstring);
wolffd@0 21 error(errstring);
wolffd@0 22 end
wolffd@0 23
wolffd@0 24 if net.nwts ~= length(w)
wolffd@0 25 error('Invalid weight vector length')
wolffd@0 26 end
wolffd@0 27
wolffd@0 28 nin = net.nin;
wolffd@0 29 nhidden = net.nhidden;
wolffd@0 30 nout = net.nout;
wolffd@0 31
wolffd@0 32 mark1 = nin*nhidden;
wolffd@0 33 net.w1 = reshape(w(1:mark1), nin, nhidden);
wolffd@0 34 mark2 = mark1 + nhidden;
wolffd@0 35 net.b1 = reshape(w(mark1 + 1: mark2), 1, nhidden);
wolffd@0 36 mark3 = mark2 + nhidden*nout;
wolffd@0 37 net.w2 = reshape(w(mark2 + 1: mark3), nhidden, nout);
wolffd@0 38 mark4 = mark3 + nout;
wolffd@0 39 net.b2 = reshape(w(mark3 + 1: mark4), 1, nout);