AutoML

Now that you have created multiple neural network models, you understand that there are two main components that go into creating well-performing networks. They are as follows:

• The architecture of the neural network
• The hyperparameters of the neural network

Depending on the problem, it could take tens of iterations to get to the best possible network. So far, we have been creating architectures and tuning the hyperparameters manually. AutoML can help us perform these tasks. It searches for the most optimal network and parameters for the dataset at hand. Auto-Keras is an open source library that helps us implement AutoML on Keras. Let's learn about how to use Auto-Keras with the help of an exercise.

Exercise 59: Get a Well-Performing Network Using Auto-Keras

In this exercise, we will make use of the Auto-Keras library to find the most optimal network and parameters for the cats-vs-dogs dataset (https://github.com/TrainingByPackt/Data-Science-with-Python/tree/master/Chapter08...