Deeplearning: Public example, "Simple Regression of Simple Functions" do not work

KNIME Extensions
1

Hi,

The above example looks awsome.
However, when executing, the output of the "Prediction" node is Nan for all of the predicted columns.

It seem to ocurr on two different KNIME installations,
( KNIME Analytics Platform    3.3.1.v201612192115
  KNIME Deeplearning4J Integration (64bit only)    3.3.0.v201612012034)
Also, re-installed the deep learning extension with no results

Could you assist?

Many thanks in advance

Moran

2

Hi gmoran,

sorry for your troubles. I looked at the workflow and somehow the configuration of the used network and Learner got messed up. If you change the activation function of both Dense Layers to 'TanH' and set the Global Learning Rate of the Learner node to '0.01' it should work fine.

Regards

David

3

Thanks David
Working nicely after applying the change.

4

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.