The Complete SPRLIB & ANNLIB

bp_adapt_weights

- adapt weights marked ``changed'' according to the accumulated delta's

SYNOPSIS

int bp_adapt_weights (net, eta, alpha, options)

ARGUMENTS

NET *net A pointer to a NET.
double eta The coefficient eta (the learning rate) of the backpropagation rule.
double alpha The coefficient alpha (the momentum term) of the backpropagation rule.
long options See the note in bp_adapt_unit.

RETURNS

TRUE if an error was detected, FALSE if no error was detected.

FUNCTION

This function is called only for shared weights networks, since for this type of network first all the gradients have to be known before they can be summed to calculate the actual weight update. Therefore, during the training of shared weights networks, bp_adapt_unit is called for each unit in a certain layer followed by one call to this function, and so on to the first layer.

NOTE

In the options-parameter the following flags are to be specified:
HISTU (Don't) store history of unit values at each update.
HISTT (Don't) store history of unit thetas at each update.
HISTW (Don't) store history of weights at each update.
BPACCUM (Don't) accumulate the delta's.
BPUPDATE (Don't) update the weights.

See bp_adapt_unit for an explanation of the possible flag combinations. This function is used by bp_adapt_net.

SEE ALSO

bp_adapt_net, bp_learn

This document was generated using api2html on Thu Mar 5 09:00:00 MET DST 1998