|
Abstract:
|
In functional data classification, functional observations are often contaminated by various systematic effects, such as random batch effects caused by device artifacts, or fixed effects caused by sample-related factors. Another issue of concern is the selection of functions when the predictors consist of multiple functions with redundancy, which we call ``curve selection." In this paper, we use a Bayesian hierarchical model to take into account random batch effects, and propose a block-wise variable selection method to choose effective curves among multiple functional predictors. Fixed effects or predictors in non-functional form are also included in the model. A hybrid Metropolis-Hastings/Gibbs sampler is used for posterior sampling, and an Evolutionary Monte Carlo algorithm is applied to improve the mixing property of the curve selection parameter.
|
