Convolutional Normalizing Flows

Understand

Bayesian posterior inference is prevalent in various machine learning problems.

• Variational inference provides one way to approximate the posterior distribution, however its expressive power is limited and so is the accuracy of resulting approximation.
• Recently, there has a trend of using neural networks to approximate the variational posterior distribution due to the flexibility of neural network architecture.
• One way to construct flexible variational distribution is to warp a simple density into a complex by normalizing flows, where the resulting density can be analytically evaluated.