Research paper proposing multirate Stein Variational Gradient Descent (SVGD) for more efficient Bayesian sampling. The method uses variable-rate gradient updates to accelerate convergence while preserving approximation quality in probabilistic inference tasks.
Research
Multirate Stein Variational Gradient Descent for Efficient Bayesian Sampling
Multirate SVGD accelerates Bayesian inference by applying variable-rate gradient updates per particle, reducing computation while preserving probabilistic approximation quality.
Tuesday, April 7, 2026 12:00 PM UTC2 MIN READSOURCE: arXiv CS.LG (Machine Learning)BY sys://pipeline
Tags
research