arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1807.02191stat.ME
收藏

通过经验过程进行经验贝叶斯推断和贝叶斯敏感性分析的MCMC方法

An MCMC Approach to Empirical Bayes Inference and Bayesian Sensitivity Analysis via Empirical Processes

Hani Doss, Yeonhee Park

摘要:

考虑一个贝叶斯情形,我们观察$Y \sim p_θ$,其中$θ\in Θ$,我们有一个关于$Θ$的潜在先验分布族$\{ ν_h, \, h \in \mathcal{H} \}$。令$g$是$θ$的实值函数,并且令$I_g(h)$是当先验是$ν_h$时$g(θ)$的后验期望。我们感兴趣的两个问题:(i)选择一个特定的$h$值,(ii)估计家庭的后验期望$\{ I_g(h), \, h \in \mathcal{H} \}$。令$m_y(h)$为超参数$h$:$m_y(h) = \int p_θ(y) \, ν_h(dθ)$的边际似然。根据定义,$h$的经验贝叶斯估计是使$m_y(h)$最大化的$h$的值。事实证明,通常可以使用马尔可夫链蒙特卡罗来形成连续体中每个个体$h$的$m_y(h)$和$I_g(h)$的点估计,以及$m_y(h)$和$I_g(h)$的有效逐点置信区间。然而,我们感兴趣的是形成估计,置信声明,整个家庭的积分$\{ m_y(h), \, h \in \mathcal{H} \}$和$\{ I_g(h), \, h \in \mathcal{H} \}$:我们需要估计的第一个家庭,以便进行经验贝叶斯推断,我们需要估计的第二个家庭,以便做贝叶斯敏感性分析。我们建立强一致性和功能的中心极限定理,这些家庭的估计,使用的工具,从经验过程理论。我们给出了两个应用程序,一个潜在的Dirichlet分配,这是用于主题建模,另一个是一个模型的贝叶斯变量选择线性回归。

英文摘要:

Consider a Bayesian situation in which we observe $Y \sim p_θ$, where $θ\in Θ$, and we have a family $\{ ν_h, \, h \in \mathcal{H} \}$ of potential prior distributions on $Θ$. Let $g$ be a real-valued function of $θ$, and let $I_g(h)$ be the posterior expectation of $g(θ)$ when the prior is $ν_h$. We are interested in two problems: (i) selecting a particular value of $h$, and (ii) estimating the family of posterior expectations $\{ I_g(h), \, h \in \mathcal{H} \}$. Let $m_y(h)$ be the marginal likelihood of the hyperparameter $h$: $m_y(h) = \int p_θ(y) \, ν_h(dθ)$. The empirical Bayes estimate of $h$ is, by definition, the value of $h$ that maximizes $m_y(h)$. It turns out that it is typically possible to use Markov chain Monte Carlo to form point estimates for $m_y(h)$ and $I_g(h)$ for each individual $h$ in a continuum, and also confidence intervals for $m_y(h)$ and $I_g(h)$ that are valid pointwise. However, we are interested in forming estimates, with confidence statements, of the entire families of integrals $\{ m_y(h), \, h \in \mathcal{H} \}$ and $\{ I_g(h), \, h \in \mathcal{H} \}$: we need estimates of the first family in order to carry out empirical Bayes inference, and we need estimates of the second family in order to do Bayesian sensitivity analysis. We establish strong consistency and functional central limit theorems for estimates of these families by using tools from empirical process theory. We give two applications, one to Latent Dirichlet Allocation, which is used in topic modelling, and the other is to a model for Bayesian variable selection in linear regression.

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。

↑