arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2307.14973stat.MEstat.CO
收藏

不充分吉布斯采样

Insufficient Gibbs Sampling

CEREMADE, Université Paris Dauphine(巴黎第九大学 CEREMADE);Department of Statistics, University of Warwick(华威大学统计系)

Antoine Luciano, Christian P. Robert, Robin J. Ryder

AI总结:

针对隐私场景下仅能获取稳健低效统计量、无法得到完整数据的问题,提出基于Gibbs采样与潜在增广数据的参数后验采样方法,还可借此估计贝叶斯因子,经示例与真实收入数据验证并明确了局限性。

摘要:

在一些应用场景中,完整数据的获取受到限制,这通常是出于隐私考量;人们只能访问从数据中推导得出的聚合型、稳健且低效的统计量。这些稳健统计量不具备充分性,但由于其具有更高的崩溃点,因此对异常值的敏感性更低,且能提供更强的数据保护。我们考虑一个参数化框架,并提出一种方法,用于在给定各类稳健且低效的统计量(具体为(中位数, MAD)或(中位数, IQR)对,或一组分位数)的条件下,从参数的后验分布中采样。我们的方法利用Gibbs采样器并模拟潜在增广数据,这有助于从属于特定分布族的参数的后验分布中进行模拟。这些基于观测统计量得到的参数与数据的联合后验分布样本的一个副产品是,我们可以通过桥接采样基于观测统计量估计贝叶斯因子。我们通过简单示例以及一项对真实收入数据的应用,验证了所提方法的有效性并概述了其局限性。

英文摘要:

In some applied scenarios, the availability of complete data is restricted, often due to privacy concerns; only aggregated, robust and inefficient statistics derived from the data are made accessible. These robust statistics are not sufficient, but they demonstrate reduced sensitivity to outliers and offer enhanced data protection due to their higher breakdown point. We consider a parametric framework and propose a method to sample from the posterior distribution of parameters conditioned on various robust and inefficient statistics: specifically, the pairs (median, MAD) or (median, IQR), or a collection of quantiles. Our approach leverages a Gibbs sampler and simulates latent augmented data, which facilitates simulation from the posterior distribution of parameters belonging to specific families of distributions. A by-product of these samples from the joint posterior distribution of parameters and data given the observed statistics is that we can estimate Bayes factors based on observed statistics via bridge sampling. We validate and outline the limitations of the proposed methods through toy examples and an application to real-world income data.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑