Variational Inference
Turns posterior approximation into optimization: pick a tractable family q, then minimize the KL divergence from q to the posterior by maximizing the evidence lower bound (ELBO) with gradient methods. Far faster than MCMC but biased — it typically underestimates posterior variance.
This Concept is waiting for its first lesson!
Turns posterior approximation into optimization: pick a tractable family q, then minimize the KL divergence from q to the posterior by maximizing the evidence lower bound (ELBO) with gradient methods. Far faster than MCMC but biased — it typically underestimates posterior variance.
Are you a teacher? Sign in to start contributing.
Sign In