Predictive risk estimation for the expectation maximization algorithm with Poisson data

DOI10.1088/1361-6420/ABE950MaRDI QIDQ5859746zbMATH OpenOpenAlexFDO

Authors Paolo Massa, Federico Benvenuto

Publication date 20 April 2021

Published in Inverse Problems (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/2011.05965

zbMATH Keywords

expectation maximizationimage deconvolutionKullback-Leibler divergencePoisson datapredictive riskStein's unbiased risk estimator-type estimator

Mathematics Subject Classification ID

Central limit and other weak theorems (60F05) Point estimation (62F10) Asymptotic properties of parametric estimators (62F12) Numerical solution to inverse problems in abstract spaces (65J22) Image processing (compression, reconstruction, etc.) in information and communication theory (94A08)

Abstract: In this work, we introduce a novel estimator of the predictive risk with Poisson data, when the loss function is the Kullback-Leibler divergence, in order to define a regularization parameter's choice rule for the Expectation Maximization (EM) algorithm. To this aim, we prove a Poisson counterpart of the Stein's Lemma for Gaussian variables, and from this result we derive the proposed estimator showing its analogies with the well-known Stein's Unbiased Risk Estimator valid for a quadratic loss. We prove that the proposed estimator is asymptotically unbiased with increasing number of measured counts, under certain mild conditions on the regularization method. We show that these conditions are satisfied by the EM algorithm and then we apply this estimator to select its optimal reconstruction. We present some numerical tests in the case of image deconvolution, comparing the performances of the proposed estimator with other methods available in the literature, both in the inverse crime and non-inverse crime setting.

Recommendations

• The expectation-maximization algorithm for ill-posed integral equations: a convergence analysis
• scientific article; zbMATH DE number 4159863
• Some properties of adding a smoothing step to the EM algorithm
• A discrepancy principle for the Landweber iteration based on risk minimization
• General convergent expectation maximization (EM)-type algorithms for image reconstruction

Cites work

• A discrepancy principle for Poisson data
• A discrepancy principle for the Landweber iteration based on risk minimization
• A natural identity for exponential families with applications in multiparameter estimation
• A parameter choice rule for Tikhonov regularization based on predictive risk
• A robust inversion method according to a new notion of regularization for Poisson data with an application to nanoparticle volume determination
• A study on regularization for discrete inverse problems with model-dependent noise
• Corrigendum: Efficient gradient projection methods for edge-preserving removal of Poisson noise
• Empirical risk minimization as parameter choice rule for general linear regularization methods
• Estimation of Kullback-Leibler losses for noisy recovery problems within the exponential family
• Estimation of the mean of a multivariate normal distribution
• scientific article; zbMATH DE number 1194487 (Why is no real title available?)
• scientific article; zbMATH DE number 3567782 (Why is no real title available?)
• scientific article; zbMATH DE number 936298 (Why is no real title available?)
• Inverse acoustic and electromagnetic scattering theory
• Inverse problems as statistics
• Inverse problems with Poisson data: statistical regularization theory, applications and algorithms
• Joint myopic deconvolution
• Linear inverse problems with discrete data. I. General formulation and singular system analysis
• Regularization of multiplicative iterative algorithms with nonnegative constraint
• Regularization parameter selection methods for ill-posed Poisson maximum likelihood estimation
• Risk estimators for choosing regularization parameters in ill-posed problems -- properties and limitations
• Stein Unbiased GrAdient estimator of the Risk (SUGAR) for Multiple Parameter Selection
• The EM Algorithm and Related Statistical Models
• The expectation-maximization algorithm for ill-posed integral equations: a convergence analysis
• The ML–EM algorithm in continuum: sparse measure solutions
• Univariate Discrete Distributions

Cited in

(4)

• Risk sensitive filtering with Poisson process observations
• A minimum chi‐squared method for indirect parameter estimation from Poisson data
• Reconstructing the image scanning microscopy dataset: an inverse problem
• Kernel Poisson regression machine for stochastic claims reserving

Describes a project that uses

Uses Software

• MLEM

This page was built for publication: Predictive risk estimation for the expectation maximization algorithm with Poisson data

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5859746)