Now that you've seen the basics behind convolutional networks and some common use cases for them, let's take a look at how to build one with CNTK.
We're going to build a model that can recognize handwritten digits in images. There's a free dataset available called the MNIST dataset that contains 60,000 samples of handwritten digits. There's also a test set available with 10,000 samples for the MNIST dataset.
Let's get started and see what building a convolutional network looks like in CNTK. First, we'll look at how to put together the structure of the convolutional neural network, we then will take a look at how to train the parameters of a convolutional neural network. Finally, we'll explore how to improve the neural network by changing it's structure with different layer setups.
