Title: Symmetrisation for nonparametric functional ANOVA
Authors: Adam Kashlak - University of Alberta (Canada) [presenting]
Abstract: Testing for equality of means and covariances among functional data groups has received a lot of attention from both parametric approaches via Gaussian processes and nonparametric ones reliant on permutation tests. We advance nonparametric testing by devising an exact test via a type of Khintchine inequality, a symmetrisation result for random variables in Banach spaces. This approach combines the computational speed of parametric methods with the distribution free benefits of permutation tests. The methodology is very general and can be extended to other data comparisons across categories.