SSRN Author: Mattias VillaniMattias Villani SSRN Content
http://www.ssrn.com/author=1260350
http://www.ssrn.com/rss/en-usFri, 17 Jul 2015 13:23:47 GMTeditor@ssrn.com (Editor)Fri, 17 Jul 2015 13:23:47 GMTwebmaster@ssrn.com (WebMaster)SSRN RSS Generator 1.0New: Speeding Up MCMC by Efficient Data SubsamplingThe computing time for Markov Chain Monte Carlo (MCMC) algorithms can be prohibitively large for datasets with many observations, especially when the data density for each observation is costly to evaluate. We propose a framework where the likelihood function is estimated from a random subset of the data, resulting in substantially fewer density evaluations. The data subsets are selected using an efficient Probability Proportional-to-Size (PPS) sampling scheme, where the inclusion probability of an observation is proportional to an approximation of its contribution to the log-likelihood function. Three broad classes of approximations are presented. The proposed algorithm is shown to sample from a distribution that is within <i>O</i>(<i>m</i>−½) of the true posterior, where m is the subsample size. Moreover, the constant in the <i>O</i>(<i>m</i>−½) error bound of the likelihood is shown to be small and the approximation error is demonstrated to be negligible even for a small m in our ...
http://www.ssrn.com/abstract=2592889
http://www.ssrn.com/1389418.htmlTue, 14 Apr 2015 09:18:39 GMT