Speaker：Professor Shu-Chuan Chen (Department of Mathematics and Statistics, Idaho State University)
Topic：Improving mixture tree reconstruction using likelihood-tuning approach
Speaker：Professor Shu-Chuan Chen
(Department of Mathematics and Statistics, Idaho State University)
Date Time：FRI. Jan 8, 2021, 10:40 AM - 11:30 AM
Place： 4F-427, Assembly Building I
Abstract
The ancestral mixture models, has been proposed by Chen and Lindsay for hierarchical clustering of long binary sequence data. In this talk, we will introduce the ancestral mixture models, how to use it to construct a hierarchical tree. We then propose two alternative improvements of the EM algorithm used in Chen and Lindsay (2006). One is the FixEM. It is just the regular EM but we no longer update the weights used in the ancestral mixture models. The other is the ModalEM which we cluster data according to the modes of an estimated density function for the data. We then proposes a likelihood tuning tool that enhances the nonparametric density estimator. We treat the kernel density estimator as one element of a model that consists of all mixtures of the kernel. By applying an EM step on the uniform density as a starting value, we can obtain the traditional kernel density estimator and a second EM step produces a fitted density with a higher likelihood. In order to compare these methods with each other and other popular hierarchical clustering methods, we use a data example from the international HapMap project. We compare the speed and the ability of these methods to separate out true clusters.
This is a joint work with Dr. Yeojin Chung and Dr. Bruce Lindsay.
This is a joint work with Dr. Yeojin Chung and Dr. Bruce Lindsay.
Keywords: ancestral mixture models, mutation kernel, likelihood tuning method