Throughout this book, you have learned many ML algorithms and techniques. What remains, however, is to deploy these algorithms into real-world applications. This chapter is dedicated to those pieces of advice related to using ML in the real world, in real applications, and in production environments.
There are many differences between idealized usage of ML algorithms and real-world usage. In our examples, we both train and execute models in one step, in response to one command. We assume that the models do not need to be serialized, saved, or reloaded in any way. We have not thought about user interface responsiveness, executing on mobile devices, or building API interfaces between clients and servers.
Real applications may also have a scope several orders of magnitude larger than the examples we've discussed. How do you train...