A hybrid neural variational CF-NADE for collaborative filtering using abstraction and generation
作者：
Highlights：
• Variational CF-NADE generative model for intractable density latent vectors.
• Reusing the shared weights for fast sampling and quality of ratings produced.
• User-item ratings and metadata features utilized with no independent assumptions.
• User-item interaction function learned using abstraction and generation phases.
• Experiments on MovieLens and Netflix datasets validate our approach.
摘要
•Variational CF-NADE generative model for intractable density latent vectors.•Reusing the shared weights for fast sampling and quality of ratings produced.•User-item ratings and metadata features utilized with no independent assumptions.•User-item interaction function learned using abstraction and generation phases.•Experiments on MovieLens and Netflix datasets validate our approach.
论文关键词：Metadata,Recommender system,Deep learning,Collaborative filtering,Generative model
论文评审过程：Received 15 January 2021, Revised 11 March 2021, Accepted 14 April 2021, Available online 21 April 2021, Version of Record 11 May 2021.
论文官网地址：https://doi.org/10.1016/j.eswa.2021.115047
