Datasets and libraries

Now that we have covered DBNs from a theoretical perspective, we take a look at some examples of code using the TensorFlow library along with the TensorFlow DBN Git repository (https://github.com/albertbup/deep-belief-network/). The repository allows you to develop simple, fast, Python implementations of DBN, which is based on binary RBMs.

We will consider the following two commonly used datasets in the machine learning community, in order to do so:

• MNIST dataset: For this dataset, you can refer to Chapter 3, Convolutional Neural Networks for Image Processing. This is a dataset of images, each of which displays a number from 0-9. Each image is 28 pixels in height and 28 pixels in width. It is available in the sklearn library but can be downloaded from the following web page, http://yann.lecun.com/exdb/mnist/:

• Boston house prices dataset: It contains...