Dynamics of the maximum marginal likelihood hyper-parameter estimation in image restoration : gradient descent vs. EM algorithm

Abstract

Dynamical properties of image restoration and hyper-parameter estimation are investigated by means of statistical mechanics. We introduce an exactly solvable model for image restoration and derive differential equations with respect to macroscopic quantities. From these equations, we evaluate relaxation processes of the system to the equilibrium state. Our statistical mechanical approach also enable us to investigate the hyper-parameter estimation by means of maximization of marginal likelihood by using gradient decent and EM algorithm from dynamical point of view.

latex 22 pages using revtex, 6 ps figures

Cited by in corpus (5)

• Algorithmic detectability threshold of the stochastic block model
• Statistical Mechanics of the Bayesian Image Restoration under Spatially Correlated Noise
• Dynamics of Order Parameters of Non-stoquastic Hamiltonians in the Adaptive Quantum Monte Carlo Method
• Statistical inference of co-movements of stocks during a financial crisis
• Deterministic flows of order-parameters in stochastic processes of quantum Monte Carlo method