A0780
Title: Informed weighted Dirichlet process mixture for functional clustering in highly correlated high-dimensional data
Authors: Wenyu Gao - University of North Carolina at Charlotte (United States) [presenting]
Inyoung Kim - Virginia Tech (United States)
Abstract: Functional clustering in high-dimensional data poses challenges, especially in scenarios with unknown cluster counts. While nonparametric Bayesian methods such as the Dirichlet process mixture (DPM) model offer approaches, they often do not effectively leverage observational information. Conversely, the weighted Dirichlet process mixture (WDPM) model incorporates prior information via a weight function. However, its investigation remains limited, particularly in functional clustering. Informed weight functions are explored for WDPM in functional clustering, addressing the gap in research by exploring covariates beyond Euclidean distances. The method is applied to fMRI data from autism spectrum disorder (ASD) patients, integrating spatial correlations and demographic information to enhance clustering accuracy.
