In any real job working in an AI team, one of the primary goals will be to build regression models which can make predictions in non-linear datasets. Because of the complexity of the real world and the data that you'll be working with, simple linear regression models won't provide you the predictive power you're seeking. This is why in this chapter, we will discuss on how to build a world-class predictions models using Multilayer Perceptron (MLPs) more information can be found at http://www.deeplearningbook.org/contents/mlp.html:

We will implement a neural network with a simple architecture of only 2 layers using TensorFlow that will perform regression on the MNIST dataset (http://yann.lecun.com/exdb/mnist/) which we'll provide. We can (and will) go deeper in architecture in later projects! We assume that you are already familiar with backpropagation (if not please read article on backpropagation by Michal Nielsen at http://neuralnetworksanddeeplearning.com/chap2.html). We'll not spend much time on how TensorFlow works, but you can refer to this official tutorial (https://www.tensorflow.org/versions/r0.10/get_started/basic_usage.html) if you are interested in looking under the hood on that technology.