next up previous
Next: Langevin Up: Initialization Previous: Backpropagation


Manhattan learning (eq. (gif)) is selected by setting MSTJN(5) = 1. It uses basically the same learning parameters as BP. However, since the weight update is not reduced by the derivative of the sigmoid, the learning rate must be a few orders of magnitude smaller.

System PRIVILEGED Account
Fri Feb 24 11:28:59 MET 1995