Kernels in the Space of Phylogenetic Trees

• Activity Number: 610
• Type: Contributed
• Date/Time: Thursday, August 7, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Section on Nonparametric Statistics
• Author(s): Grady Weyenberg*+
• Companies: University of Kentucky
• Keywords: phylogenetics ; classification

Abstract

In a previous paper we developed a nonparametric method for detecting unusual (or outlying) individuals in a collection of phylogenetic trees. The method uses geodesic distance in BVH tree space to define kernels, and attempts to classify the trees based on these non-normalized densities. In contrast to a more classic metric space, the volume of the kernels depends not only the selected bandwidth for the kernel, but also the combinatoric structure of the space near the sampled point. Here, we explore the structure of tree space and present some bounds for the volume of the kernels. These results are used to improve the algorithm from our previous work.