The mu factor is used to increase and decrease the step-size during quickprop training. The mu factor should always be above 1, since it would otherwise decrease the step-size when it was suppose to increase it.
The default mu factor is 1.75.
Neural network resource.
The mu factor, or false on error.