Deep Learning By Example

Summary

In this chapter, we went through the idea of representation learning and why it's useful for doing deep learning or machine learning in general on input that's not in a real-valued form. Also, we covered one of the adopted techniques for converting words into real-valued vectors—Word2Vec—which has very interesting properties. Finally, we implemented the Word2Vec model using the skip-gram architecture.

Next up, you will see the practical use of these learned representations in a sentiment analysis example, where we need to convert the input text to real-valued vectors.