Handwritten Digit Recognition

Now that we have covered the mechanisms of a neural network, let's consider a practical problem. We will classify some handwritten digit images. Assuming that training has already been completed, we will use trained parameters to implement "inference" in the neural network. This inference is also called forward propagation in a neural network.

Note

In the same way as the procedure for solving a machine learning problem (which consists of two phases, "training" and "inference"), to solve a problem using a neural network, we will use training data to train the weight parameters and then use the trained parameters while predicting to classify the input data.

MNIST Dataset

Here, we will use a set of images of handwritten digits called MNIST. MNIST is one of the most famous datasets in the field of machine learning and is used in various ways, from simple experiments to research. When you read research papers...