As a computational alternative to Markov chain Monte Carlo approaches,
variational inference (VI) is becoming more and more popular for approximating
intractable posterior distributions in large-scale Bayesian models due to its
comparable efficacy and superior efficiency. Several recen
Amortized variational inference (A-VI) is a general alternative to factorized (or mean-field) variational inference (F-VI) for approximating intractable posterior distributions, with conditions derived for achieving F-VI's optimal solution and strategies for expanding the domain of the inference function, while certain models like hidden Markov models and Gaussian processes cannot be matched by A-VI.