Objective
This project concerns approximations for the distributions of non-parametric statistics and problems in non-parametric estimation theory. This requires the analysis of limit theorems and related second-order approximations for complex estimation procedures for non-parametric models for signals, images and curves perturbed by noise. Computer-intensive testing of the approximations is necessary for implementation.
Asymptotic (sequential) minimax procedures based on e.g. stochastic search, wavelet and resampling methods are also to be developed for such complex models. These models are described by a very large number of parameters with implicit context dependencies as in images and smooth curves under the influence of additive or non-additive noise.
New methods of resampling will be developed and studied using second-order asymptotics. These methods will be consistent and unbiased for estimating the distributions of a large class of relevant non-parametric statistics for which traditional bootstrap methods fail.
Topic(s)
Call for proposal
Data not availableFunding Scheme
Data not availableCoordinator
2300 RA Leiden
Netherlands