MNIST dataset analysis

In this section, we are going to get our hands dirty by implementing a classifier for handwritten images. This kind of implementation could be considered as the Hello world! of neural networks.

MNIST is a widely used dataset for benchmarking machine learning techniques. The dataset contains a set of handwritten digits like the ones shown here:

Figure 3: Sample digits from the MNIST dataset

So, the dataset includes handwritten images and their corresponding labels as well.

In this section, we are going to train a basic model on these images and the goal will be to tell which digit is handwritten in the input images.

Also, you'll find out that we will be able to accomplish this classification task using very few lines of code, but the idea behind this implementation is to understand the basic bits and pieces for building a neural network solution...