|
|
|
Probabilistic graphical models (PGMs) have proven to perform
successfully in a wide range of scenarios that require reasoning under uncertainty.
PGMs are primarily characterized by combining wisely the benefits of
probability theory with the advantages of graph theory: while the former
provides a solid ground for handling uncertainty, the latter provides an
intuitive interface to the user.
Supervised and unsupervised classification, also known as class prediction and class discovery, respectively, are two problems that involve one or another type of uncertainty. Moreover, these problems appear in many application fields (e.g. data mining, bioinformatics, robotics). PGMs offer a sound as well as intuitive solution to them. This is a research field that has been receiving increasing attention for the last few years.
We would like to invite authors to submit their works on PGMs for supervised
and unsupervised classification for a special issue of Machine
Learning on these topics. Submissions must conform with the standards
of originality, relevance and quality of Machine
Learning, as they will be subject to the regular reviewing process. Submissions
based on work published in previous conferences must represent a significant
expansion over the prior publication.
* Topics of Interest. A non-exhaustive list of the topics of interest is as follows:
- PGMs and
extensions thereof for supervised and unsupervised classification in discrete,
continuous and mixed domains
- PGM classification
learning algorithms (structural and parametric)
- Accuracy
assessment, and model validation and interpretation in PGMs
- Feature selection
and extraction for PGMs
- (Semi)automatic
identification of the number of clusters in PGM learning
- Discretization
of continuous variables for PGM learning
- Prior knowledge
incorporation in PGM classification learning algorithms
- Synergies
between supervised and unsupervised PGM classifiers
- Applications
of PGM classifiers
* Important Dates:
- Reception
of papers: February 29, 2004
- Notification
of acceptance: June 30, 2004
- Reception
of revised versions: September 30, 2004
* Submission Instructions:
Manuscripts must follow the Machine Learning
submissions instructions, which can be found in the last pages of a sample
of the journal or in the following web page: http://www.cs.ualberta.ca/~holte/mlj/index.html
In addition to
everything stated in the standard submission guidelines, submissions to this
special issue should also do the following:
- electronic submissions should be
sent to Jose M. Peña (jmp@ifm.liu.se) with the subject “Special Issue ML:
PGMs for Classification”. The email should enclose author information: the
first author will be the primary contact unless otherwise stated.
- if
electronic submission is not possible, please submit six hard copies of
the paper to:
Iñaki Inza
Department of Computer Science and Artificial
Intelligence
P.O.Box 649
E-20080 Donostia – San Sebastián
Spain
- please check that the
editors acknowledge the reception of your electronic or hard-copy submission.
* Editors:
|
Pedro
Larrañaga (ccplamup@si.ehu.es) University
of the Basque Country, Spain |
Jose A.
Lozano (lozano@si.ehu.es) University
of the Basque Country, Spain |
|
Jose M.
Peña (jmp@ifm.liu.se) Linköping
University, Sweden |
Iñaki
Inza (inza@si.ehu.es) University
of the Basque Country, Spain |
Please address
any question to Jose M. Peña (jmp@ifm.liu.se) and/or Iñaki Inza (inza@si.ehu.es)