Evolutionary inference for functional data: using Gaussian processes on phylogenies of functional data objects
Evolutionary inference for functional data: using Gaussian processes on phylogenies of functional data objects
复制标题
DOI：
--
复制
发表时间：
2012
期刊：
影响因子：
--
通讯作者：
M. Kerr
中科院分区：
文献类型：
--
作者：
M. Kerr
本论文探讨了利用系统发育学和功能数据分析来分析连续的祖先数据，如连续曲线。高斯过程(GP)被放置在系统发展图上，以便对功能数据对象执行进化推断。GP的均值和协方差函数模拟了系统发育中不同状态之间的关系。功能数据对象完全由协方差函数内的空间和时间参数描述，从而允许例如通过最大似然估计方法进行推断。成功地对已知的系统发育、缺少祖先数据的系统发育和未知拓扑的系统发育进行了推断。这项工作对于那些想要在连续的祖先数据上计算进化推断的人来说是潜在的有用的，对于这些数据来说，系统发育GP被证明是一个有效和有前途的工具。
This thesis explores the use of phylogenetics and functional data analysis for the analysis of continuous ancestral data such as continuous curves. Gaussian processes (GPs) are placed on phylogenies in order to perform evolutionary inferences on the functional data objects. The mean and covariance functions of the GP model the relationships between different states on the phylogeny. The functional data objects are completely described by the spatial and temporal parameters within the covariance functions, allowing inferences to be made, for example, by the method of maximum likelihood estimation. Inferences are successfully made on known phylogenies, phylogenies with missing ancestral data and on phylogenies of unknown topology. This work is potentially useful for those wanting to compute evolutionary inferences on continuous ancestral data, for which phylogenetic GPs are shown to be an efficient and promising tool.