This session will be a detailed study of deep learning techniques for recommendation tasks. We will start the session by discussing matrix factorization for recommendations (e.g. for the Netflix prize). Then we will explore deep learning methods as extensions of the matrix factorization. In particular, we will discuss auto-encoders, graph embeddings, and graph neural networks. We will also talk about deep neural network architectures for recommendations. e.g. wide and deep networks, siamese/triplet loss networks, etc.
Key Takeaways for the Audience
- Building state-of-the-art recommendation systems
- Deep learning applications
- Auto-encoders for recommendations
- Graph deep learning and applications
