Variational Inference

An approximate method for Bayesian inference that transforms posterior estimation into an optimization problem by minimizing KL divergence, using a parameterized simple distribution to approximate a complex posterior distribution.