[Comp-neuro] ML-book

Norbert Jankowski norbert at is.umk.pl
Fri Nov 7 20:03:58 CET 2008


 

Call for Book Chapters: 
Meta-learning in Computational Intelligence.

 

Publisher: Springer Verlag, Series: Studies in Computational Intelligence

Editors: Włodzisław Duch, Krzysztof Grąbczewski, Norbert Jankowski


E-mail address: mlbook at is.umk.pl


 

Large CI systems, such as WEKA or Yale, contain hundreds of transformations,
pre- and post-processing, classification, approximation and visualization
methods. Constructing a good model in a difficult case requires tedious
experiments and is very hard, especially for inexperienced users. There is
an urgent need to create systems that can learn at meta-level, discovering
the best model for a given data and eliminating the need for human
intervention in search for good models. 

 

Relevant book topics include, but are not limited to:

 

*        Criteria and methods for automated search for best data models.

*        Extraction of knowledge guiding the meta-learning process,
collection and re-use of information for learning. 

*        Strategies for search in parameter space in which data models are
specified. 

*        Methods of discovering nontrivial solutions in reasonable time. 

*        Control and selection of optimal model complexity depending on
various criteria. 

*        Meta-learning architectures: What is necessary to facilitate
meta-search from the information exchange point of view? 

*        Methods and data structures of abstract information exchange in
meta-learning. How to build, exploit and rearrange meta-knowledge? 

*        Testing methods to estimate model's quality and to direct further
learning. 

*        All meta-learning computational intelligence approaches are
welcome: statistical, machine learning, pattern recognition, data mining,
neural, evolutionary etc.

 

The chapters of this book are not devoted to applications of already known
algorithms or development of single algorithms. The proposed book will
provide clear directions and the much-needed focus for meta-learning
research, showing a roadmap towards a new generation of data mining tools.
The book is mainly intended for people engaged in Computational Intelligence
and related research fields, either in their theoretical or practical
aspects. 

 

Submission procedure: Meta-learning researchers are invited to submit an
extended abstract (1-2 pages) before Dec 30, 2008. The abstracts will be
reviewed by the editors with respect to the book topics, and the authors of
selected abstracts will be invited to submit full chapters before Apr 3,
2009. 

The chapters should have about 20-25 pages (in the printed book). The
typesetting should be done in LATEX with the help of a standard Springer
LATEX package. The chapters will be reviewed by Computational Intelligence
research experts and the revised versions of the accepted chapters will be
expected before Jun 15, 2009. 

We expect the book to be printed in late Summer 2009.

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.neuroinf.org/pipermail/comp-neuro/attachments/20081107/dd4eb78e/attachment.html


More information about the Comp-neuro mailing list