bayesian-embedding
Bayeisan Embedding is about introducing latent variables to enhance or manipulate the priors of embedding to serve following reasons
- Enhance explanibility
- If there's finite set of different latent only in concern
Literatures
Bayeisan Skip Grams (BSG)
Embedding Words as Distributions with a Bayesian Skip-gram Model, Arthur Bražinskas, Serhii Havrylov, Ivan Titov https://arxiv.org/pdf/1711.11027.pdf https://github.com/ixlan/BSG
Literatures about bayesian-embedding + graph
Bayes EMbedding (BEM): Refining Representation by Integrating Knowledge Graphs and Behavior-specific Networks https://arxiv.org/pdf/1908.10611.pdf