wolffd@0: wolffd@0:
wolffd@0:wolffd@0: net = glminit(net, prior) wolffd@0:wolffd@0: wolffd@0: wolffd@0:
net = glminit(net, prior)
takes a generalized linear model
wolffd@0: net
and sets the weights and biases by sampling from a Gaussian
wolffd@0: distribution. If prior
is a scalar, then all of the parameters
wolffd@0: (weights and biases) are sampled from a single isotropic Gaussian with
wolffd@0: inverse variance equal to prior
. If prior
is a data
wolffd@0: structure similar to that in mlpprior
but for a single layer of
wolffd@0: weights, then the parameters
wolffd@0: are sampled from multiple Gaussians according to their groupings
wolffd@0: (defined by the index
field) with corresponding variances
wolffd@0: (defined by the alpha
field).
wolffd@0:
wolffd@0:
glm
, glmpak
, glmunpak
, mlpinit
, mlpprior
Copyright (c) Ian T Nabney (1996-9) wolffd@0: wolffd@0: wolffd@0: wolffd@0: