LSSD 2014 - Special Session on 'Learning of structured and non-standard data'

Special Session on
'Learning of structured and non-standard data'
23-25 April 2014, Bruges, Belgium
http://www.dice.ucl.ac.be/esann
AIMS AND SCOPE
Today real life data are often given not in the form of vectorial
data but in various formats often without an underlying metric space.
Prominent examples are network structured data e.g. from social or
communication networks, tree structured data used to represent hierarchical
documents or collections thereof. Also the simple relation between
objects e.g. by means of score data found in sequence alignments
or rating information as obtained in collaborative filtering approaches
is of this type.
The recent technological developments, also in the context of big data,
allow the generation of very complex data sets.
Challenges are now in the effective processing of these data,
in the light of the pure amount, but also to keep the obtained
model informative for subsequent analysis. These data are also
often given without an explicit vector space, point relations
can be asymmetric, metric properties may not be valid and the
available information if often sparse in different representations.
Computational intelligence methods have the potential to be used to pre-process,
model and to analyze such complex data but new strategies are needed.
A very effective way is to employ explicit or implicit knowledge about the data,
or the analysis task and to learn an appropriate model from available
training data.
In other cases the knowledge is used in the design of adaptive
analysis algorithms
to generate the desired meta information out of the data.
Such knowledge may be available by means of appropriate (bio-)physical models,
data specific distance measures, auxiliary information associated with the data.
or dedicated processing strategies for non-metric data employing
infinite kernels
or dissimilarity learning approaches.
Also novel data encoding techniques and projection methods, employing
concepts from randomization algorithm, have been used to obtain compact
descriptions of these complex data sets or to identify relevant information.
Examples of such data analysis problems are e.g. in the analysis of
biological or social networks with a large number of measurements
and complex data relations.
TOPICS
We encourage submission of papers on novel methods for structured data,
dissimilarity learning, non-standard data analysis and non-metric data
processing
by means of computational intelligence and machine learning approaches,
including but not limited to:
- data analysis and pattern recognition approaches for structured data
- dissimilarity learning
- methods employing ex- and implicit data knowledge for non-standard data
- representation and modeling of heterogeneous, high-dimensional,
multi-modal (structured) and/or non-standard data
- approaches in the line of matrix completion, collaborative filtering,
reduction techniques for non-standard data
- large scale network analysis
IMPORTANT DATES
Paper submission deadline : 29 November 2013
Notification of acceptance : 31 January 2014
Deadline for final papers : 21 February 2014
The ESANN 2014 conference : 23-25 April 2014
SPECIAL SESSION ORGANIZERS:
Frank-Michael Schleif, University of Appl. Sc. Mittweida, Germany and
University of Birmingham, Birmingham, UK
Thomas Villmann, University of Appl. Sc. Mittweida, Germany
Peter Tino, University of Birmingham, Birmingham, UK