Summary
Sequential data is one of the most active research fields in deep learning, especially because natural language data is sequential. But sequential data processing is not just limited to that. Time series data, which is essentially everything that happens around us, including sound, other waveforms, and more, is all sequential in nature.
The most difficult problem in processing sequence data is long-term dependency, but sequential data comes with a lot more complexities. RNNs were the breakthrough in the sequence data processing field. Thousands of different variations of RNNs have been explored by researchers and it's still an actively growing field.
In this chapter, we have gone through the basic building blocks of sequential data processing. Although we have worked only with the English language, the techniques we have learned here are generally applicable to any type of data. Understanding these building blocks is crucial for beginners because everything that ...
