Summary
In this chapter, we have covered three different topics. We started with a summary of the many options for reading, converting, and writing neural networks.
We then moved on to the deployment of neural networks, using the sentiment analysis case study from Chapter 7, Implementing NLP Applications, as an example. The goal here was to build a workflow that uses the trained neural network to predict the sentiment of new reviews stored in the database. We have shown that a deployment workflow can be assembled in two ways: manually or automatically with Integrated Deployment.
The last section of the chapter dealt with the scalability of network training and execution. In particular, it showed how to exploit the computational power of GPUs when training a neural network.
In the next and last chapter of this book, we will explore further deployment options and best practices when working with deep learning.
