The MNIST dataset

The Modified National Institute of Standards and Technology (MNIST) is a large database of handwritten digits. It has a set of 70,000 examples of data. It is a subset of NIST's larger dataset. The digits are of 28x28 pixel resolution and are stored in a matrix of 70,000 rows and 785 columns; 784 columns form each pixel value from the 28x28 matrix, and one value is the actual digit. The digits have been size-normalized and centered in a fixed-size image.

The digit images in the MNIST set were originally selected and experimented with by Chris Burges and Corinna Cortes using boundingbox normalization and centering. Yann LeCun's version uses centering by center of mass within a larger window. The data is available on Yann LeCun's ...

Get Keras 2.x Projects now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.