Cross-validation for a Keras Neural Network

Hello,
I would like to add a k fold cross-validation (stratified) to my neural network made with Keras. My final activation fonction is a sigmoid (one output neuron, for a binary classification). Do you have any idea on how to build it. I tried to use the x-partitioner(and the agregator) by affecting the classes according to the probabilities, but it doesn’t work. Do you have any idea ?
I’m still a beginner, any help would be highly appreciated.
Thank you in advance !

Hi @Soufiane8413,

Welcome to the KNIME community! :slight_smile:

I don’t see any obvious reasons why you can’t replace the Learner and Predictor nodes in https://kni.me/w/ph4V4jcg5pNTRZfb with Keras equivalents.

What exactly have you tried and what was the error? Could you share a workflow that demonstrates the issue together with some sample data? This would be extremely helpful in figuring out what’s going wrong.

Best,
Stefan

2 Likes

Hello @stelfrich. I’ve get this workflow https://kni.me/w/ph4V4jcg5pNTRZfb and it now works perfectly ! I don’t even know the reason ^^ Thank you very much !

1 Like

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