Sequence data is data where the order matters, such as in audio, video, and speech. Learning sequential data is one of the most challenging problems in the field of pattern recognition because of the nature of the data. The dependencies between the parts of sequences and their varying length add further complexity when processing sequential data. With the advent of sequence models and algorithms such as recurrent neural networks (RNN), long short-term memory models (LSTM), and gated recurrent units (GRU), sequence data modeling is being utilized in multiple applications, such as sequence classification, sequence generation, speech to text conversion, and many more.

In sequence classification, the goal is to predict the category of the sequence, whereas in sequence generation, we generate a new output sequence based on the input sequence. In this chapter, you will learn how to implement sequence classification and generation, as well as time series forecasting, using different flavors of RNN. 

In this chapter, we will cover the following recipes:

• Sentiment classification using RNNs
• Text generation using LSTMs
• Time series forecasting using GRUs
• Implementing bidirectional recurrent neural networks