Overcoming Posterior Collapse in Variational Autoencoders Via EM-Type Training.
Ying LiLei ChengFeng YinMichael Minyi ZhangSergios TheodoridisPublished in: ICASSP (2023)
Keyphrases
- image segmentation
- probabilistic model
- feedforward neural networks
- restricted boltzmann machine
- denoising
- expectation maximization
- bayesian framework
- probability distribution
- test set
- training examples
- em algorithm
- machine learning
- training set
- generative model
- training samples
- maximum likelihood
- optical flow
- gaussian mixture model
- posterior distribution
- training algorithm
- image processing
- deep learning
- computer vision