Having established the fundamentals of TensorFlow tensors and operations, we now turn to constructing machine learning models. TensorFlow integrates Keras as its high-level Application Programming Interface (API), which simplifies the process of defining neural network architectures. This approach allows you to focus more on the model design rather than low-level operational details.

In this chapter, you will learn to:

• Recognize the role of Keras in the TensorFlow ecosystem for efficient model development.
• Build models composed of linear stacks of layers using the Keras Sequential API.
• Define more intricate model structures, including those with multiple inputs or outputs, using the Functional API.
• Identify and use common layer types (like Dense, Conv2D, Dropout) and activation functions (such as ReLU, Sigmoid, Softmax) as building blocks.
• Understand the basics of creating custom layers and models for specialized requirements.

We will conclude by applying these concepts to construct a basic classification model, giving you practical experience with the Keras workflow from definition to instantiation.