Chapter 9. Deep learning for sequences and text
- How sequential data differs from nonsequential data
- Which deep-learning techniques are suitable for problems that involve sequential data
- How to represent text data in deep learning, including with one-hot encoding, multi-hot encoding, and word embedding
- What RNNs are and why they are suitable for sequential problems
- What 1D convolution is and why it is an attractive alternative to RNNs
- The unique properties of sequence-to-sequence tasks and how to use the attention mechanism to solve them
9.1. Second attempt at weather prediction: Introducing RNNs
9.2. Building deep-learning models for text
9.3. Sequence-to-sequence tasks with attention mechanism
Materials for further reading