The demo features a feedforward sigmoid neural network that can recognize handwritten numeric digits.
The network comprises a structure of 785, 30, and 10 neurons. The process involves scaling down the bounding box of the drawing to
20x20 and centering it on the center of mass of the pixels in a 28x28 image. This image is then fed into the pre-trained neural
network according to the instructions for how the MNIST training data was made.