RNN

Created: 2022-03-24 10:09

Very useful for sequential data processing, i.e. temporal dynamics or patterns -> good in case we want to consider temporal evolution of user's interests

Consider variants like LSTM or GRU for vanishing gradient problem.

Good for session in which user is not logged in -> not our case

Some works imply that a simple neighbourhood approach could reach same accuracy as some RNN models -> see this discussion

Doubts

References

1. https://medium.com/sciforce/deep-learning-based-recommender-systems-b61a5ddd5456

Tags

#rnn