The Complete SPRLIB & ANNLIB

PTRON-flags

- flags for perceptron learning

SYNOPSIS

FIXED_UPD / BOUND_UPD Learn with a constant learning rate / a learning rate which decreases with a fixed step per iteration cycle.
NO_NORM_UPD / NORM_UPD Do not normalize / normalize the weight vectors during the learning process.
SYNC_UPD / ASYNC_UPD Update the weights each time after showing the entire set / after showing a sample (comparable to batch update vs. sample update in back-propagation - see BP-flags).

NOTE

The pairs of flags are mutually exclusive. The flags can be OR'ed together to fill the mode parameter in the PTRON structure, which controls perceptron learning.

SEE ALSO

PTRON, learn_ptron

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