Improving Marginal Likelihood Estimation for Bayesian Phylogenetic Model Selection
Improving Marginal Likelihood Estimation for Bayesian Phylogenetic Model Selection
复制标题
DOI：
10.1093/sysbio/syq085
复制
发表时间：
2011-03-01
影响因子：
6.5
通讯作者：
Chen, Ming-Hui
中科院分区：
文献类型：
--
作者：
Xie, Wangang;Lewis, Paul O.;Chen, Ming-Hui
边际似然通常用于比较贝叶斯系统发生学中的不同进化模型，并且是计算贝叶斯因子以比较模型拟合度时使用的中心量。调和平均 (HM) 方法是一种流行的估计边际似然的方法，可以根据马尔可夫链蒙特卡罗分析的输出轻松计算，但通常会大大高估边际似然。热力学积分（TI）方法比 HM 方法准确得多，但需要更多的计算。在本文中，我们介绍了一种新方法，垫脚石采样（SS），它使用重要性采样来估计桥接后验分布和先验分布的一系列比率（“垫脚石”）中的每个比率。我们在模拟和使用真实数据中比较了 SS 方法与 TI 和 HM 方法的性能。我们得出的结论是，尽管需要额外的计算，但 SS 和 TI 方法的准确性大大提高，因此值得使用它们来代替 HM 方法。
The marginal likelihood is commonly used for comparing different evolutionary models in Bayesian phylogenetics and is the central quantity used in computing Bayes Factors for comparing model fit. A popular method for estimating marginal likelihoods, the harmonic mean (HM) method, can be easily computed from the output of a Markov chain Monte Carlo analysis but often greatly overestimates the marginal likelihood. The thermodynamic integration (TI) method is much more accurate than the HM method but requires more computation. In this paper, we introduce a new method, stepping-stone sampling (SS), which uses importance sampling to estimate each ratio in a series (the "stepping stones") bridging the posterior and prior distributions. We compare the performance of the SS approach to the TI and HM methods in simulation and using real data. We conclude that the greatly increased accuracy of the SS and TI methods argues for their use instead of the HM method, despite the extra computation needed.