The Amortized Bootstrap
نویسندگان
چکیده
We use amortized inference in conjunction with implicit models to approximate the bootstrap distribution over model parameters. We call this the amortized bootstrap, as statistical strength is shared across dataset replicates through a metamodel. At test time, we can then perform amortized bagging by drawing multiple samples from the implicit model. We find amortized bagging outperforms bagging with independent parameter estimates across a variety of prediction tasks.
منابع مشابه
Statistical Topology Using the Nonparametric Density Estimation and Bootstrap Algorithm
This paper presents approximate confidence intervals for each function of parameters in a Banach space based on a bootstrap algorithm. We apply kernel density approach to estimate the persistence landscape. In addition, we evaluate the quality distribution function estimator of random variables using integrated mean square error (IMSE). The results of simulation studies show a significant impro...
متن کاملSemiparametric Bootstrap Prediction Intervals in time Series
One of the main goals of studying the time series is estimation of prediction interval based on an observed sample path of the process. In recent years, different semiparametric bootstrap methods have been proposed to find the prediction intervals without any assumption of error distribution. In semiparametric bootstrap methods, a linear process is approximated by an autoregressive process. The...
متن کاملBootstrap confidence intervals of CNpk for type‑II generalized log‑logistic distribution
This paper deals with construction of confidence intervals for process capability index using bootstrap method (proposed by Chen and Pearn in Qual Reliab Eng Int 13(6):355–360, 1997) by applying simulation technique. It is assumed that the quality characteristic follows type-II generalized log-logistic distribution introduced by Rosaiah et al. in Int J Agric Stat Sci 4(2):283–292, (2008). Discu...
متن کاملA New Bootstrap Based Algorithm for Hotelling’s T2 Multivariate Control Chart
Normality is a common assumption for many quality control charts. One should expect misleading results once this assumption is violated. In order to avoid this pitfall, we need to evaluate this assumption prior to the use of control charts which require normality assumption. However, in certain cases either this assumption is overlooked or it is hard to check. Robust control charts and bootstra...
متن کاملOptimum Block Size in Separate Block Bootstrap to Estimate the Variance of Sample Mean for Lattice Data
The statistical analysis of spatial data is usually done under Gaussian assumption for the underlying random field model. When this assumption is not satisfied, block bootstrap methods can be used to analyze spatial data. One of the crucial problems in this setting is specifying the block sizes. In this paper, we present asymptotic optimal block size for separate block bootstrap to estimate the...
متن کامل