CNN on Keras with Tensorflow backend

August 27, 2017

Report Abuse
This sample shows that we can import Tensorflow as the backend for Keras into Azure ML Studio for usage in Execute Python Script.
Here you will not see how to save trained model or how to use pre-trained models. As ML Studio for an output of Execute Python Script requires a dataset to be a pandas.DataFrame as an alternative to autosaving model by callback you can try to write your manual callback which will get and save the weights, and then transform that array to pandas.DataFrame for the output of Python Script.