Here's how to get MNIST, from the sklearn docs: >>> from sklearn.datasets import fetch_mldata >>> mnist = fetch_mldata('MNIST original', data_home=custom_data

Here's how to get MNIST, from the sklearn docs: >>> from sklearn.datasets import fetch_mldata >>> mnist = fetch_mldata('MNIST original', data_home=custom_data_home) I think the handwritten digits dataset used in the presentation is just a subset of MNIST; MNIST is 28x28 and the handwritten digits are 8x8.