Deep Learning with R Cookbook

There's more...

Google's Cloud Machine Learning (CloudML) Engine also provides an automated tool for model hyperparameter tuning. It allows us to test different hyperparameter configurations when training a model. To submit a hyperparameter tuning job, we need to pass a CloudML training configuration file to the cloudml_train() function. We discussed hyperparameter tuning using Keras in the There's more... section of the Training your first deep neural network recipe in Chapter 1, Understanding Neural Networks and Deep Neural Networks, where we showcased how to tune model parameters. We first defined flags for parameters that we want to optimize and then used these flags in the definition of our model.

In the same way, we define the flags ...