Description of data files

MNIST

is a famous hand-written digit data set. There are different variants of it, but this is derived from an original data set often
used in competitions.

mnist_train.csv

The file contains the 60,000 training examples and labels.

mnist_test.csv

The file contains 10,000 test examples and labels.

Each row in both files consists of 794 values: first, 784 pixel values (a number from 0 to 255), then, the label in a 1-of-10 encoding (10 digits).
Hence, the image size is 28x28. Note that the order of values is different to the original files.

mnist_train.xpat

The file contains the data of 'mnist-train.csv' in the compressed Boone pattern format. Instead of a single label, there is the
classical classification encoding of the target class.

mnist_test.xpat

The file contains the data of 'mnist-test.csv' in the compressed Boone pattern format with the same modifications as found
in the training set.