Machine Learning in Elixir

Wrapping Up

In this chapter, you were introduced to natural language processing with recurrent neural networks. You learned how to implement some basic text featurization and preprocessing with Elixir and how to implement and train recurrent neural networks in Axon. You also learned about some of the intuition behind recurrent neural networks and some of the fundamental shortcomings of recurrent neural networks, which have allowed them to be usurped by transformer models.

If you’re familiar with the current state of natural language processing, you might have questioned the need to cover recurrent neural networks at all. However, to better understand the current state of the art, it’s important to understand the previous state of the art—and ...