Efficient Marginalization of Discrete and Structured Latent Variables via Sparsity.
Gonçalo M. CorreiaVlad NiculaeWilker AzizAndré F. T. MartinsPublished in: NeurIPS (2020)
Keyphrases
- latent variables
- posterior distribution
- probabilistic model
- gaussian process
- structured output
- latent variable models
- topic models
- approximate inference
- prior knowledge
- random variables
- variational bayes
- real valued
- hierarchical model
- hidden variables
- observed variables
- conditionally independent
- factor graphs
- machine learning
- structured prediction
- causal relationships
- gaussian processes
- message passing
- image reconstruction
- probability distribution
- high dimensional