Estimation and annealing for Gibbsian fields
Consider a Gibbs distribution \(\pi_{\theta}(x)=\exp (-<\alpha (x),\theta >)Z(\theta)^{-1}\) on a finite configuration space. The maximum likelihood estimator \(\theta_*\) is given as the solution of \(E_{\theta}[\alpha]=\alpha (x)\). A direct evaluation of \(E_{\theta}[\alpha]\) or a calculation of \(\theta_*\) by the Robbins- Monro algorithm is not feasible here because no simple simulation method of \(\pi_{\theta}\) is available. The very nice main idea of the paper is to use a stochastic gradient algorithm with the so-called Gibbs-sampler which is an inhomogeneous Markov chain \(P_{\theta}^{n,n+1}\) converging weakly to \(\pi_{\theta}\). It is shown hat the sequence of iterates \(\theta\) (n) of this algorithm converges asymptotically to \(\theta_*\) if the step size is not too small. Practical aspects of implementing this algorithm are also discussed. Finally it is shown that the annealing algorithm continues to work with \(\theta_*\) replaced by \(\theta\) (n), i.e. \(P^{n,n+1}_{\theta (n)/T(n)}\) converges weakly to the uniform distribution on \(\{x| <\alpha (x),\theta_*>\) is minimal\(\}\) if T(n)\(\to 0\) slowly. These results have applications in image processing.
- Parametric inference for imperfectly observed Gibbsian fields
- Convergence of some partially parallel Gibbs samplers with annealing
- Asymptotic behaviour of time-inhomogeneous evolutions on von Neumann algebras
- Simulated annealing with time-dependent energy function
- Outperforming the Gibbs sampler empirical estimator for nearest-neighbor random fields
- Long range search for maximum likelihood in exponential families
- Convergence and convergence rate of stochastic gradient search in the case of multiple and non-isolated extrema
- Asymptotics of maximum likelihood estimators based on Markov chain Monte Carlo methods
- On the theoretical properties of the exchange algorithm
- Monte Carlo likelihood inference for missing data models
- A general stochastic model for algorithms of distribution estimation with conditional probabilities and Gibbs sampling
- Learning undirected graphical models using persistent sequential Monte Carlo
- Some recent research in the analysis of mixture distributions
- scientific article; zbMATH DE number 5158941
- scientific article; zbMATH DE number 4007524 (Why is no real title available?)
- scientific article; zbMATH DE number 1208108 (Why is no real title available?)
- Stochastic approximation algorithms for partition function estimation of Gibbs random fields
- Markow chain markov field dynamics:models and statistics
- Bayesian inference in the presence of intractable normalizing functions
- scientific article; zbMATH DE number 219007 (Why is no real title available?)
- Bayesian computation for statistical models with intractable normalizing constants
- Gibbs Flow for Approximate Transport with Applications to Bayesian Computation
- Stochastic adaptive selection of weights in the simulated tempering algorithm
- Convergence of Markovian stochastic approximation for Markov random fields with hidden variables
- Marginal problem, statistical estimation, and Möbius formula
- An ergodic L2-theorem for simulated annealing in bayesian image reconstruction
- Segmentation of macular fluorescein angiographies. A statistical approach
- The S-U algorithm for missing data problems
- Hyperparameter estimation for satellite image restoration using a MCMC maximum-likelihood method
- An algorithm for estimating the unknown parameter of the Gibbs distribution based on the stochastic quasi-gradient method
- A computational framework for empirical Bayes inference
- Hidden Markov random field models applied to color homogeneity evaluation in dyed textile images
- Logarithmic integral optimization via adaptive importance sampling based surrogation methods
- Simulated annealing and quantum detailed balance
This page was built for publication: Estimation and annealing for Gibbsian fields
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1106598)
