Hands-On Deep Learning with R

Creating user and item embeddings

Recommender systems can use deep neural networks to support complex, non-linear data representations in flexible, scalable, and efficient ways.

Embeddings are low-dimensional representations (vectors) of continuous numbers learned from representations (vectors) of discrete input variables in neural networks. As previously noted in this chapter, recommender systems typically need an index of similarity between users and user-item preferences to identify similar users and find ratings (preferences) of unrated items.

However, unlike traditional collaborative filtering approaches that use generalized matrix factorization methods to produce user-item affinity vectors, neural networks can store important information ...