Hebbian Learning Utility Process

Description:

Self normalising Hebbian learning rule.

State data:

alpha (DOUBLE) (1x1):

Hebbian learning rate parameter.

norm (DOUBLE) (0, 1, 2) (default = 1):

Self normalisation. 0 = none, 1 = over afferents, 2 = over efferents.

savemode (STRING) ('read' 'overwrite' 'safewrite' 'append' 'safeappend'):

The savemode sets how data will be stored at the end of a run and read at the start. 'read' 'append' and 'safeappend' require the weights to already have been saved. 'overwrite' starts with fresh weights and then saves the final weights over the last saved weight file. 'append' starts with the last saved weights file and then saves the final weights over the last saved weight file. The 'safe' modes save to a new file with an incrementing filename at the end of the runs. 'read' reads back the last run and does no learning. No weights are saved at the end of the run.

savename (STRING) ('??'):

The name of the weight save file.