next up previous
Next: Conjugate Gradient Up: Initialization Previous: Langevin


Quickprop ( MSTJN(5)=3), which estimates the updating step through eq. (gif), has two learning parameters and two control parameters. These are the ``learning rate'' PARJN(1) (), the sigmoid-prime addition PARJN(23) (), the maximum growth factor PARJN(21) and the maximum weight magnitude PARJN(22). The latter is quite unimportant. Default values recommended in [9] for the other three parameters are; PARJN(1) of order unity, PARJN(23)= 0.1 and PARJN(21) = 1.75. We have not done extensive tests to to determine the problem dependence of these parameters.

It is recommended to run QP using a small weight decay ( PARJN(5)).

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