momentum - trains the network using momentum descent


    momentum [<num-updates>] [-report <report-interval> | -setOnly]


This is a shortcut for training the network using momentum descent. The arguments are similar to those for train. The momentum term is taken from the network's momentum parameter.

Momentum descent incorporates a percentage (determined by the network's momentum parameter) of the previous step in each weight change. This can cause the effective learning rate to grow as high as 1 / (1 - momentum).

If the -setOnly flag is used, no training will occur. However, the network's numUpdates, reportInterval, and default algorithm will be set. This can be used to set the default training behavior in an initialization script prior to actually training.


To train for 1000 epochs, printing reports every 100:

    lens> momentum 1000 -r 100


train, steepest, dougsMomentum, deltaBarDelta

Last modified: Fri Nov 17 12:36:39 EST 2000