Online Program

Return to main conference page
Saturday, June 1
Computational Statistics
Computational Statistics E-Posters
Sat, Jun 1, 9:30 AM - 10:30 AM
Grand Ballroom Foyer

Wavelet Shrinkage Using Bayesian False Discovery Rate Methods: a Comparison Study (306256)


*Rodney Vasconcelos Fonseca, Unicamp 
Debashis Mondal, Oregon State University 

Keywords: Empirical Bayes, False discovery rate, Wavelet shrinkage

Wavelet methods consist in a popular tool for function denoising that has been largely investigated. An interesting idea consists to handle the process of wavelet shrinkage as multiple hypothesis tests on the wavelet coefficients, which can be performed applying the concept of false discovery rate (FDR) control. However, this technique has not received much attention in the literature, and basically two papers provide the main ways to follow this idea, using the Bayesian approach to wavelet shrinkage as foundation. In this work we compare these two Bayesian approaches to apply the FDR method with wavelet coefficients and also propose a novel way of applying them through empirical Bayes methods. Application of the discussed techniques are illustrated with both simulated and real data.