Coreset Markov Chain Monte Carlo

arXiv:2310.17063v2 Announce Type: replace-cross
Abstract: A Bayesian coreset is a small, weighted subset of data that replaces the full dataset during inference in order to reduce computational cost. However, state of the art methods for tuning coreset weights are expensive, require nontrivial user input, and impose constraints on the model. In this work, we propose a new method -- Coreset MCMC -- that simulates a Markov chain targeting the coreset posterior, while simultaneously updating the coreset weights using those same …

abstract art arxiv bayesian computational constraints cost data dataset however inference markov reduce small stat.co state state of the art stat.ml type work