What is the connection between Gradient Descent and variational inference?

Gradient Descent is often used to optimize variational parameters in variational inference frameworks.

Variational inference transforms Bayesian inference into an optimization problem by minimizing KL divergence between an approximate distribution and the true posterior. Gradient Descent is used to optimize the parameters of the variational distribution. This connects probabilistic modeling with deterministic optimization, enabling scalable Bayesian learning in large models where exact inference is intractable.