Tuning and Deploying Deep Neural Networks
So far in the journey of this book, we have primarily talked about how to develop a DNN for a given use case and looked at a few strategies and rules of thumb to bypass roadblocks we could face in the process. In this chapter, we will discuss the journey onward after developing the initial model by exploring the methods and approaches you need to implement when the model developed doesn’t perform to your expectations. We will discuss regularization and hyperparameter tuning, and toward the end of the chapter, we will also have a high-level view of the process to deploy a model after tuning. However, we won’t actually discuss the implementation specifics of deploying; this will just be an overview offering guidelines to achieve success in the process. Let’s get started.