Recurrent Neural Networks
Recurrent Neural Networks (RNNs) are the primary modern approach for modeling data that is sequential in nature. The word "recurrent" in the name of the architecture class refers to the fact that the output of the current step becomes the input to the next one (and potentially further ones as well). At each element in the sequence, the model considers both the current input and what it "remembers" about the preceding elements.
Natural Language Processing (NLP) tasks are one of the primary areas of application for RNNs: if you are reading through this very sentence, you are picking up the context of each word from the words that came before it. NLP models based on RNNs can build on this approach to achieve generative tasks, such as novel text creation, as well as predictive ones such as sentiment classification or machine translation.
In this chapter, we'll cover the following topics:
- Text generation ...
