Pain Point #1: Importing MNIST images

One of the most common datasets used for image classification is the MNIST dataset, which is composed of thousands of samples of handwritten digits. The Modified National Institute of Standards and Technology (MNIST) is, according to Yann LeCun, Corinna Cortes, and Christopher J.C. Burges, useful for the following reasons:

Note

It is a good database for people who want to try learning techniques and pattern recognition methods on real-world data while spending minimal efforts on preprocessing and formatting.

There are several methods to import the MNIST images into our Jupyter notebook. We will cover the following two methods in this chapter:

1. Directly through the TensorFlow library
2. Manually through the MNIST website

Note

One thing to note is that we will be primarily using MNIST images as our example of how to improve performance within a convolutional neural network. All of these techniques that will be applied on MNIST images can be applied to any image that...