In this chapter, you learned how to utilize deep learning algorithms for practical applications. The fields that are well studied are image recognition and NLP. While learning about the field of NLP, we looked through two new deep learning models: the RNN and LSTM networks, which can be trained with time sequential data. The training algorithm used in these models is BPTT. You also learned that there are three approaches to make the best of the deep learning ability: the field-oriented approach, the breakdown-oriented approach, and the output-oriented approach. Each approach has a different angle, and can maximize the possibility for deep learning.
And …congratulations! You've just accomplished the learning part of deep learning with Java. Although there are still some models that have not been mentioned yet in this book, you can be sure there will be no problem in acquiring and utilizing them. The next chapter will introduce some libraries that are implemented with other programming...