Collaborative filtering using deep learning

There are numerous strategies for performing collaborative filtering using neural networks. These can be roughly grouped into two categories: retrieval-based and sequential.

Retrieval-based CF models

The most common models assign a probability that a user will ever be interested in a given item. These items are then ranked, and the top can be selected. Examples include two-tower models, GNNs, and neural collaborative filtering.

Sequential CF models

Other models attempt to predict the items with which the user will next interact with (where often ). Examples include transformers and LSTM models.