Markov Chain Monte-Carlo (MCMC)

Markov Chain Monte-Carlo (MCMC) is used as an alternative to Variational Inference, by sampling from distribution.

The Monte Carlo part of the algorithm refers to random sampling. Monte-Carlo methods are a class of methods based on repeated random sampling. A Markov Chain is a mathematical model which governs the behavior between a sequence of events. Each node in the chain represents a state and are joined together by links which is the transition probability between states.