The Complete SPRLIB & ANNLIB

samann_learn: - perform a number of learning cycles with the backpropagation learning algorithm

SYNOPSIS

int samann_learn

(net, data_set, eta, alpha, cycles, options)

ARGUMENTS

`NET`	*`net`**	Pointer to a net; must be initialised by `samann_init`.
`DATASET`	*`data_set`**	A pointer to a learning set (must be an `IOSET).`
`double`	`eta`	The coefficient eta (the learning rate) of the SAMANN rule.
`double`	`alpha`	The coefficient alpha (the momentum term) of the SAMANN rule.
`long`	`cycles`	Number of learning cycles to be performed.
`long`	`options`	See the note below.

RETURNS

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

FUNCTION

After some initial checking, the function samann_learn_sample is called cycles times. There are two possible definitions of a cycle. If the SARNDPICK flag is set in options, one cycle means picking two samples randomly from the dataset and passing them to the samann_learn_sample function. Otherwise, one cycle means taking all possible combinations of samples in set and passing them to samann_learn_sample. Note that this can computationally become very intensive, since there are 1/2 * n * (n-1) combinations of n samples. In the latter case, if the flag SAVERBOSE is set in options, a counter is printed to stdout indicating which pair of samples is currently considered. For a full description of the update rule, please see: Mao, J. and Jain, A.K., Artificial neural networks for feature extraction and multivariate data projection, IEEE Transactions on Neural Networks, 6(2):296-317, 1995.

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.
`SAACCUM`	(Don't) accumulate the delta's.
`SAUPDATE`	(Don't) update the weights.
`SARNDPICK`	(Don't) update one sample-pair at a time.
`SAVERBOSE`	(Don't) print a counter when updating the units.

Note that the HIST-flags and SAMANN-flags may be ORed. Furthermore, note the following for (SAACCUM, SAUPDATE) - triples:

0 0 0: Combination does not make sense (nothing is done).

0 0 1: Yields no backpropagation, only updates.

0 1 0: Yields only backpropagation, no updates.

0 1 1: Specifies normal mode, i.e. backpropagation and update.

The Complete SPRLIB & ANNLIB

`samann_learn`

- perform a number of learning cycles with the backpropagation learning algorithm

SYNOPSIS

ARGUMENTS

RETURNS

FUNCTION

NOTE

SEE ALSO

`0`	`0`	0: Combination does not make sense (nothing is done).
`0`	`0`	1: Yields no backpropagation, only updates.
`0`	`1`	0: Yields only backpropagation, no updates.
`0`	`1`	1: Specifies normal mode, i.e. backpropagation and update.