A0641
Title: Model data heterogeneity with Dirichlet diffusion trees
Authors: Shuning Huo - Amazon (United States)
Hongxiao Zhu - Virginia Tech (United States) [presenting]
Abstract: A challenge of modern data analysis is the difficulty to model complex data heterogeneity structures caused by sub-populations or latent factors. We propose a Bayesian latent tree model to characterize data heterogeneity and link the heterogeneity structure with covariates. We adopt Dirichlet Diffusion Trees to model the latent hierarchical data structure underlying the observed data, and propose a regression framework by associating covariates with the parameters of the latent trees. To perform posterior inference, we propose a Markov chain Monte Carlo algorithm to alternatively update the latent tree structures and the regression coefficients. We demonstrate the effectiveness of the model through a simulation study and imaging data on brain Glioblastoma Multiforme images.
