Forward Amortized Inference for Likelihood-Free Variational Marginalization.
Luca AmbrogioniUmut GüçlüJulia BerezutskayaEva W. P. van den BorneYagmur GüçlütürkMax HinneEric MarisMarcel van GervenPublished in: AISTATS (2019)
Keyphrases
- message passing
- sum product
- sum product algorithm
- image segmentation
- maximum likelihood
- search tree
- worst case
- probabilistic inference
- dirichlet process mixture models
- bayesian inference
- bayesian networks
- posterior probability
- variational methods
- forward and backward
- methods in computer vision
- search algorithm
- belief networks
- image restoration
- random fields
- distributed systems
- inference process
- free energy
- optical flow