wolffd@0: wolffd@0:
wolffd@0:wolffd@0: wolffd@0: g = rbfgrad(net, x, t) wolffd@0: [g, gdata, gprior] = rbfgrad(net, x, t) wolffd@0:wolffd@0: wolffd@0: wolffd@0:
g = rbfgrad(net, x, t)
takes a network data structure net
wolffd@0: together with a matrix x
of input
wolffd@0: vectors and a matrix t
of target vectors, and evaluates the gradient
wolffd@0: g
of the error function with respect to the network weights (i.e.
wolffd@0: including the hidden unit parameters). The error
wolffd@0: function is sum of squares.
wolffd@0: Each row of x
corresponds to one
wolffd@0: input vector and each row of t
contains the corresponding target vector.
wolffd@0: If the output function is 'neuroscale'
then the gradient is only
wolffd@0: computed for the output layer weights and biases.
wolffd@0:
wolffd@0: [g, gdata, gprior] = rbfgrad(net, x, t)
also returns separately
wolffd@0: the data and prior contributions to the gradient. In the case of
wolffd@0: multiple groups in the prior, gprior
is a matrix with a row
wolffd@0: for each group and a column for each weight parameter.
wolffd@0:
wolffd@0:
rbf
, rbffwd
, rbferr
, rbfpak
, rbfunpak
, rbfbkp
Copyright (c) Ian T Nabney (1996-9) wolffd@0: wolffd@0: wolffd@0: wolffd@0: