Summary

In this chapter, we learned to set up Datalab to execute neural networks on Google Cloud. We also learned the structure of a neural network and how various parameters, such as depth, number of hidden units, activation function, optimizer, batch size, and number of epochs, impact the accuracy of the model. We also saw how to implement a neural network in both Keras and TensorFlow. Topics such as using premade estimators and creating custom estimators in TensorFlow were covered.