You are viewing a plain text version of this content. The canonical link for it is here.

Posted to user@mahout.apache.org by Karol Grzegorczyk <kg...@agh.edu.pl> on 2014/05/23 10:09:46 UTC

Multiple epoch MLP training

Hi,

Typically, MLP is being trained using backpropagation algorithm run in 
many epochs. Running backprop in only one epoch is not enough to build a 
reasonable model. Unfortunately, I do not see the number of epoch as an 
input parameter to the MLP training job in Mahout. Is it a deliberate 
decision or it is only due to the immaturity of the implementation?

Regards,
Karol