Applying the basic multilayer perceptron on classifying number in the MNIST dataset.
Number of inputs - 784 - based on the 28x28 pixel image size.
Number of outputs - 10 - one for each class, i.e the numbers from 1 to 10.
No hidden layers.
Training set - 50 000. Validation set - 10 000. Test set - 10 000.
Implemented with Python.