MNIST handwritten digit database

This is a database for handwritten digit classification, used in the Deep Learning chapter 18.
Each greyscale image is 28 x 28, representing the digits 0-9.
This dataset is large, consisting of 60,000 training images and 10,000 test images.

We do not reproduce the dataset here, but point to our source:

provided by Yann Le Cun, a well known neural network expert.