TensorFlow is an open source software library created by Google that allows you to build and execute data flow graphs for numerical computation. In these graphs, every node represents some computation or function to be executed, and the graph edges connecting up nodes represent the data flowing between them. In TensorFlow, the data is multi-dimensional arrays called Tensors. Tensors flow around the graph, hence the name TensorFlow.

Machine learning (ML) models, such as convolutional neural networks, can be represented with these kinds of graphs, and this is exactly what TensorFlow was originally designed for.

In this chapter, we'll cover the following topics:

• Understanding the TensorFlow way of thinking
• Setting up and installing TensorFlow
• Introduction to TensorFlow API levels
• Building and training a linear classifier in TensorFlow
• Evaluating a trained model