In this chapter, you learned how to validate different types of deep learning models and how you can use metrics in CNTK to implement validation logic for your models. We also explored how to use TensorBoard to visualize training progress and the structure of the model so you can easily debug your models.
Monitoring and validating your model early and often will ensure that you end up with neural networks that work very well on production and do what your client expects them to. It is the only way to detect underfitting and overfitting of your model.
Now that you know how to build and validate basic neural networks, we'll dive into more interesting deep learning scenarios. In the next chapter, we will explore how you can use images with neural networks to perform image detection, and in Chapter 6, Working with Time Series Data, we will take a look at how to build...
