MATLAB Answers

Best practice for data normalization when using import model from Keras to predict ?

3 views (last 30 days)
Tuong Nguyen Minh
Tuong Nguyen Minh on 1 Sep 2020
Answered: Madhav Thakker on 15 Sep 2020
I have train a deep learning model in Keras where my input data is normalize by using the standard scaler
If I import that model into matlab to use it to predict I also have to use the standard scaler.
Now I have two choices in mind to do this
1/ Call the standard scaler from python
2/ Use Matlab built in function
My Matlab version is 2020a
What choice should I proceed with and how to implement it
Thank you very much

Answers (1)

Madhav Thakker
Madhav Thakker on 15 Sep 2020
Hi Tuong,
I understand that you want to do the preprocessing as done with the training data by using StandardScaler of sklearn. The same functionality can be done with MATLAB zscore. To replicate the behavior of the scipy function, you will have to set flag input as 1, to use population standard deviation.
Hope this helps.

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!