In this chapter, you learned how to implement deep learning models with the libraries ND4J and DL4J. Both support GPU computing and both give us the ability to implement them without any difficulties. ND4J is a library for scientific computing and enables vectorization, which makes it easier to implement a calculation among arrays because we don't need to write iterations within them. Since machine learning and deep learning algorithms have many equations with vector calculations, such as inner products and element-wise multiplication, ND4J also helps implement them.

DL4J is a library for deep learning, and by following some examples with the library, you saw that we can easily build, train, and evaluate various types of deep learning models. Additionally, while building the model, you learned why regularization is necessary to get better results. You also got to know some optimizers of the learning rate: momentum, ADAGRAD, and ADADELTA. All of these can be implemented easily with DL4J.

You gained knowledge of the core theories and implementations of deep learning algorithms and you now know how to implement them with little difficulty. We can say that we've completed the theoretical part of this book. Therefore, in the next chapter, we'll look at how deep learning algorithms are adapted to practical applications first and then look into other possible fields and ideas to apply the algorithms.