Suggestion by Hib Engler: add a function net_grow that doubles the number of neurons in the hidden layer(s), but keeps the input/output relation between the input and output layer.
The version in CVS has this functionality already.
Log in to post a comment.
The version in CVS has this functionality already.