Reliable amortized variational inference with physics-based latent distribution correction.
Ali SiahkoohiGabrio RizzutiRafael OrozcoFelix J. HerrmannPublished in: CoRR (2022)
Keyphrases
- variational inference
- exponential family
- bayesian inference
- latent variables
- posterior distribution
- gaussian process
- variational methods
- topic models
- probabilistic model
- probabilistic graphical models
- closed form
- mixture model
- worst case
- probability distribution
- random variables
- latent dirichlet allocation
- density estimation
- graphical models
- log likelihood
- information retrieval
- maximum likelihood
- information extraction
- upper bound
- factor graphs
- feature selection