I would like to have a keras model self-contained to reduce the training/serving skew.

It would mean here having a preprocessing layer that is doing essentially what MinMaxScaler from scikit learn is doing to have data ranging from 0 to 1.

I know about the normalization layer, but I dont want to normalize my inputs but to scale it to 0-1 interval.

Rescaling layer is closer to what I need but it seems like it requires all the inputs features being on the same scale at the beginning (like all features being between 0 and 255 for images). However, I have got features on different scales.

How can I add that kind of preprocessing layer in my model to have the preprocessing within the model?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.