9. Modern Recurrent Neural Networks

We have introduced the basics of RNNs, which can better handle sequence data. For demonstration, we implemented RNN-based language models on text data. However, such techniques may not be sufficient for practitioners when they face a wide range of sequence learning problems nowadays.

For instance, a notable issue in practice is the numerical instability of RNNs. Although we have applied implementation tricks such as gradient clipping, this issue can be alleviated further with more sophisticated designs of sequence models. Specifically, gated RNNs are much more common in practice. We will begin by introducing two of such widely-used networks, namely gated recurrent units (GRUs) and long short-term memory (LSTM). Furthermore, we will expand the RNN architecture with a single undirectional hidden layer that has been discussed so far. We will describe deep architectures with multiple hidden layers, and discuss the bidirectional design with both forward and backward recurrent computations. Such expansions are frequently adopted in modern recurrent networks. When explaining these RNN variants, we continue to consider the same language modeling problem introduced in Section 8.

In fact, language modeling reveals only a small fraction of what sequence learning is capable of. In a variety of sequence learning problems, such as automatic speech recognition, text to speech, and machine translation, both inputs and outputs are sequences of arbitrary length. To explain how to fit this type of data, we will take machine translation as an example, and introduce the encoder-decoder architecture based on RNNs and beam search for sequence generation.