Connectionists: CFP: Special Session at IJCNN, Interpretable models

[Michael Biehl]

First call for papers:

*Special Session at IJCNN 2017 <http://www.ijcnn.org/>*

*Interpretable models in machine learning for advanced data analysis*

*Anchorage, Alaska, USA, May 2017*

Organizers / contact: Michael Biehl (m.biehl at rug.nl), Thomas Villmann (
villmann at hs-mittweida.de)

Technological progress leads to a tremendous growth of the amount of
digital data in virtually all scientific and engineering disciplines. At
the same time, the structural complexity of the acquired data is increasing
steadily. As a consequence, it is instrumental to develop efficient methods
for automated data analysis. However, good performance of the methods in
terms of, for instance, classification or clustering is frequently not
sufficient. Very often, deeper insight into the data processing and the
problem at hand is desirable. For example, classifiers should be
interpretable as to how the classification of a particular observation is
obtained and which of the available information constitutes the basis of
the decision. These additional properties of data processing methods can be
summarized best by the term interpretability. The aim of the special
session is to present and discuss new approaches for data analysis in terms
of interpretable models, i.e. aiming at their added value beyond the mere
clustering or classification itself. Interpretability of models is
essential in nearly all areas of machine learning and data analysis. Hence,
the topic of the session should be relevant for a large variety of research
areas within the IJCNN community.

Possible topics include, but are not restricted to:
- prototype based models for unsupervised and supervised learning
- analysis of interpretable data structures
- interpretable feature extraction for improved performance
- visualization of multi-dimensional data for knowledge extraction
- integration of prior and expert knowledge
- interpretable adaptive (dis-)similarities and relevance learning

We encourage researchers interested in the theory and/or real world
applications of interpretable models to contribute to the session.
Theoretical models should be illustrated, whenever possible. Application
oriented contributions should demonstrate how the interpretable models
provide new, relevant insights into the data beyond the original task of,
e.g., classification, prediction, or clustering.

Please visit the conference homepage <http://www.ijcnn.org/> for practical
information and submission guidelines.

Important dates:
Paper submission: November 15, 2016
Decision notification: January 20, 2017
Final version due: February 20, 2017
IJCNN conference: May 14-19, 2017


----------------------------------------------------------

Prof. Dr. Michael Biehl
Johann Bernoulli Institute for
Mathematics and Computer Science
P.O. Box 407, 9700 AK Groningen
The Netherlands

Tel. +31 50 363 3997

www.cs.rug.nl/~biehl
m.biehl at rug.nl
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/connectionists/attachments/20161011/f84f507e/attachment.html>