Dear all,

We are inviting submissions for a special issue of METRON (
https://link.springer.com/journal/40300) dealing with "Hidden Markov
Models: theory and applications". Hidden Markov models (HMMs) and their
extensions (such as hidden semi-Markov models), although known for decades,
have become increasingly popular in recent years, yet in many ways are
still in a state of development. The application of HMMs has regularly been
justified by their versatility and mathematical tractability: all moments
are available in closed form; the likelihood computation is linear in the
number of observations; the marginal distributions are easy to determine;
missing observations can be handled with minimal effort; the conditional
distributions such as forecast distributions are available; and outlier
identification is possible.


As dependent mixture models, HMMs can be regarded as classifiers in a time
series context, though their use is by no means limited to such supervised
learning tasks. For example, they are natural models to adjust for
unobserved or latent heterogeneity, can be used in an unsupervised learning
context for likelihood-based clustering, and as time series models of
course also allow for forecasts. However, HMMs also experience several
technical difficulties. The likelihood may not be bounded, and, even if it
is, local maxima often exist, and the global maximum might not always be
the best choice. Algorithmic solutions are nearly always required and
algorithms such as the EM algorithm are experiencing numerous problems: for
example the appropriate choice of initial values or using an adequate
stopping rule. Model selection, including the selection of a suitable
number of states, adds one more topic to the many areas of interest. There
is also a growing body of work on regression models that are driven by
underlying states.


Topics of interest include, but are not limited to, the following:

• Algorithms

• Testing in Hidden Markov Models

• Identifiability Problems

• Multivariate Hidden Markov Models

• Nonstandard Dependence Structures in Hidden Markov Models

• Robustness of Hidden Markov Model Estimation

• Hidden Markov Models for Clustering

• Hidden Markov Models of Generalized Linear and/or Additive Models

• Bayesian Approaches for Hidden Markov Models

• Missing Data Analysis in Hidden Markov Models

• High-dimensional Hidden Markov Models.


The papers should have a methodological or advanced data analytic component
to be considered for publication. Authors who are uncertain about the
suitability of their papers should contact the special issue editors. All
submissions must contain original unpublished work not being considered for
publication elsewhere. Submissions will be refereed according to standard
procedures for METRON. Information about the journal can be found at
https://link.springer.com/journal/40300. The deadline for submissions is 30
November 2018. However, papers can be submitted at any time; and, when they
have been received, they will enter the editorial system immediately.
Papers for the special issue should be submitted using the Editorial
Manager Electronic Submission tool: http://www.editorialmanager.com/tron.
Please choose the special issue on Hidden Markov Models and one of the
Co-Editors responsible for the special issues. The special issue editors:


• Antonello Maruotti

Libera Università Maria Ss Assunta

Email: [log in to unmask]

• Jan Bulla

University of Bergen

Email: [log in to unmask]

• Roland Langrock

Bielefeld University

Email: [log in to unmask]

-- 
 
<http://www.lumsa.it/ricerca_5permille?utm_source=banner&utm_medium=email&utm_campaign=bannermail>

You may leave the list at any time by sending the command

SIGNOFF allstat

to [log in to unmask], leaving the subject line blank.