6. Conclusion

This chapter provided an overview of the three deep learning models – MLP, RNN, CNN – and also introduced TensorFlow 2 tf.keras, a library for rapid development, training, and testing deep learning models that is suitable for a production environment. The Sequential API of Keras was also discussed. In the next chapter, the Functional API will be presented, which will enable us to build more complex models specifically for advanced deep neural networks.

This chapter also reviewed the important concepts of deep learning such as optimization, regularization, and loss functions. For ease of understanding, these concepts were presented in the context of MNIST digit classification.

Different solutions to MNIST digit classification using artificial neural networks, specifically MLP, CNN, and RNN, which are important building blocks of deep neural networks, were also discussed together with their performance measures.

With an understanding of deep learning concepts and how Keras can be used as a tool with them, we are now equipped to analyze advanced deep learning models. After discussing the Functional API in the next chapter, we'll move on to the implementation of popular deep learning models. Subsequent chapters will discuss selected advanced topics such as autoregressive models (autoencoder, GAN, VAE), deep reinforcement learning, object detection and segmentation, and unsupervised learning using mutual information. The accompanying Keras code implementations will play an important role in understanding these topics.