arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2201.08502stat.ME
收藏

使用 Ellipsoid-Gaussian 分布进行曲线因子分析

Curved factor analysis with the Ellipsoid-Gaussian distribution

Hanyu Song, David B. Dunson

AI总结:

针对多元 Gaussian 分布难以建模非线性关系的问题,本文提出由球面 von Mises-Fisher 潜变量因子模型导出的 Ellipsoid-Gaussian 分布,并设计 Bayesian 混合采样算法,验证其曲线低维结构建模能力。

摘要:

人们需要用于刻画多元数据中相关性的新模型。多元 Gaussian 分布虽被常规使用,却无法刻画数据中的非线性关系。大多数非线性扩展往往高度复杂,例如需要在潜变量中估计非线性回归模型。本文提出一类相对简单的 Ellipsoid-Gaussian 多元分布,它通过一个 Gaussian 线性因子模型导出,其中潜变量在单位超球面上服从 von Mises-Fisher 分布。我们表明,Ellipsoid-Gaussian 分布能够灵活地建模具有低维结构的变量间曲线关系。采用 Bayesian 方法,我们提出一种将基于梯度的 geodesic Monte Carlo 与 adaptive Metropolis 相结合的混合方法用于后验采样。我们推导了其基本性质,并在多种模拟和真实数据应用中展示了 Ellipsoid-Gaussian 分布的效用。同时还提供了一个配套的 R 包。

英文摘要:

There is a need for new models for characterizing dependence in multivariate data. The multivariate Gaussian distribution is routinely used, but cannot characterize nonlinear relationships in the data. Most non-linear extensions tend to be highly complex; for example, involving estimation of a non-linear regression model in latent variables. In this article, we propose a relatively simple class of Ellipsoid-Gaussian multivariate distributions, which are derived by using a Gaussian linear factor model involving latent variables having a von Mises-Fisher distribution on a unit hyper-sphere. We show that the Ellipsoid-Gaussian distribution can flexibly model curved relationships among variables with lower-dimensional structures. Taking a Bayesian approach, we propose a hybrid of gradient-based geodesic Monte Carlo and adaptive Metropolis for posterior sampling. We derive basic properties and illustrate the utility of the Ellipsoid-Gaussian distribution on a variety of simulated and real data applications. An accompanying R package is also available.

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。

↑