Abstract Maximum likelihood estimation of discrete latent variable (DLV) models is usually performed by the expectation-maximization (EM) algorithm. A well-known drawback related to multimodality log-likelihood function so that algorithm can converge a local maximum, not corresponding global one. We propose tempered EM explore parameter space adequately for two main classes DLV models, namely c...

In this paper, we study the autoregressive (AR) model with normal inverse Gaussian (NIG) innovations. The NIG distribution is semi heavy-tailed and helpful in capturing extreme observations present data. expectation-maximization (EM) algorithm used to estimate parameters of considered AR(p) model. efficacy estimation procedure shown on simulated data for AR(2) AR(1) models. A comparative presen...

The Expectation-Maximization Algorithm (EM) is a widely used method allowing to estimate the maximum likelihood of models involving latent variables. When Expectation step cannot be computed easily, one can use stochastic versions EM such as Stochastic Approximation EM. This algorithm, however, has drawback require joint belong curved exponential family. To overcome this problem, [16] introduce...