Introduction

In the previous chapter, we experimented with different neural network architectures. We were able to evaluate the performance of the different models by observing the loss and accuracy during the course of the training process. This helped us determine when the model was underfitting or overfitting the training data and how to use techniques such as early stopping to prevent overfitting.

In this chapter, you will learn about cross-validation. This is a resampling technique that leads to a very accurate and robust estimation of a model's performance, in comparison to the model evaluation approaches we discussed in the previous chapters.

This chapter starts with an in-depth discussion about why we need to use cross-validation for model evaluation, the underlying basics of cross-validation, its variations, and a comparison between them. Next, we will implement cross-validation on Keras deep learning models. We will also use Keras wrappers with scikit-learn to...