Training a neural network efficiently with minibatches

In the previous chapter, we discussed how to build and train a neural network. In this chapter, we'll discuss various ways to feed data to the CNTK trainer. Before we dive into the details of each data processing method, let's take a closer look at what happens with the data when you train a neural network.

You need a couple of things to train a neural network. As we discussed in the previous chapter, you need to have a basic structure for your model and a loss function. The trainer and the learner are the final pieces to the puzzle and are responsible for controlling the training process.

The trainer performs four steps:

1. It takes a number of training samples and feeds them through the network and loss function
2. Next, it takes the output of the loss function and feeds it through the learner
3. It then uses the learner...