Bayesian wavelet shrinkage with logistic prior

Alex Rodrigo dos Santos Sousa

DOI: https://doi.org/10.1080/03610918.2020.1747076

2020-04-01

Abstract:Consider the nonparametric curve estimation problem. Wavelet shrinkage methods are applied to the data in the wavelet domain for noise reduction. After denoising, the function can be estimated by wavelet basis expansion. The present paper proposes a Bayesian approach for wavelet shrinkage with the use of a mixture of a point mass function at zero and the logistic distribution symmetric around zero as the prior distribution to the wavelet coefficients of an unknown function in models with additive Gaussian errors. Statistical properties such as bias, classical and Bayesian risks of the rules are analyzed and performances of the proposed rules are obtained in simulations studies involving the Donoho-Johnstone test functions. Application in a mass spectrometry real data set is done.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to effectively reduce noise and improve the accuracy of function estimation in non - parametric curve estimation using the Bayesian wavelet shrinkage method. Specifically, the author proposes a new Bayesian wavelet shrinkage method, which uses a mixture of the zero - mass function and the logistic distribution symmetric about zero as the prior distribution to handle the wavelet coefficients of the unknown function. This method aims to achieve effective shrinkage of the wavelet coefficients by adjusting the hyper - parameters of the prior distribution, thereby retaining the main features of the signal while denoising. The main contributions of the paper are as follows: 1. **Selection of prior distribution**: Proposing the use of the logistic distribution as the prior distribution of wavelet coefficients, which helps to better control the degree of shrinkage and can explicitly give an approximate formula for the shrinkage rule. 2. **Performance analysis**: Through simulation studies and practical data applications, the performance of the proposed Bayesian shrinkage rule under different signal - to - noise ratios is verified. The results show that this method is superior to existing shrinkage techniques in most cases. 3. **Practical application**: Applying this method to mass spectrometry data sets shows its denoising effect in practical data, especially performing well in the analysis of phospholipid composition in samples with low progesterone concentrations. In conclusion, this paper provides an effective method for noise reduction in non - parametric curve estimation by introducing a new prior distribution and detailed performance analysis.

Bayesian wavelet shrinkage with logistic prior

Kurtosis control in wavelet shrinkage with generalized secant hyperbolic prior

On posterior distribution of Bayesian wavelet thresholding

Bayesian wavelet de-noising with the caravan prior

Shrinkage with Robustness: Log-Adjusted Priors for Sparse Signals

Bayesian nonparametric regression using complex wavelets

Wavelet methods in statistics: Some recent developments and their applications

Development and Application of an Optimised Bayesian Shrinkage Prior for Spectroscopic Biomedical Diagnostics

Tail-adaptive Bayesian shrinkage

Horseshoe priors for edge-preserving linear Bayesian inversion

Bayesian singular value regularization via a cumulative shrinkage process

Controlling the flexibility of non-Gaussian processes through shrinkage priors

Bayesian Regression Using a Prior on the Model Fit: The R2-D2 Shrinkage Prior

Bayesian shrinkage prediction for the regression problem

Bayesian Analysis for Over-parameterized Linear Model without Sparsity

Penalty parameter selection and asymmetry corrections to Laplace approximations in Bayesian P-splines models

Wasserstein metric-driven Bayesian inversion with applications to signal processing

Bayesian Compressive Sensing Using Reweighted Laplace Priors

Bayesian estimation for mean vector of multivariate normal distribution on the linear and nonlinear exponential balanced loss based on wavelet decomposition

Shrinkage priors for isotonic probability vectors and binary data modeling, with applications to dose–response modeling