Summary
In this chapter, we were introduced to practical deep learning with TensorFlow 2 and Keras, their key features and applications, and how they work together. We became familiar with the differences between low- and high-level APIs, as well as how to leverage the most advanced modules to ease the creation of deep models. Then, we discussed how to implement a deep neural network with TensorFlow and addressed some major topics: from model creation, training, validation, and testing, we highlighted the most important aspects to consider so as to avoid pitfalls. We saw how to build different types of deep learning models, such as fully connected, convolutional, and recurrent neural networks, via the Keras API. We solved a regression task and a classification problem, which gave us hands-on experience with this. We learned how to leverage TensorBoard to visualize many different training trends regarding metrics and model parameters. Finally, we built and trained a model that is able to classify fashion item images with high accuracy, an activity that shows that a possible real-world problem can be solved with the help of the most advanced deep learning techniques.
In the next chapter, we will be studying the OpenAI Gym environment and how to use TensorFlow 2 for reinforcement learning.