Connectionists: NIPS Workshop on Kernel Methods and Structured Domains

[Craig Saunders]

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

                    Call for Papers

   Workshop on Kernel Methods and Structured Domains

         http://nips2005.kyb.tuebingen.mpg.de/

                       NIPS 2005

     Submission deadline:        21 October
     Accept/Reject notification: 05 November

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%


Workshop Description
%%%%%%%%%%%%%%%%%%%%

Substantial recent work in machine learning has focused on the problem 
of dealing with inputs and outputs on more complex domains than are 
provided for in the classical regression/classification setting. 
Structured representations can give a more informative view of input 
domains, which is crucial for the development of successful learning 
algorithms: application areas include determining protein structure and 
protein-protein interaction; part-of-speech tagging; the organisation of 
web documents into hierarchies; and image segmentation. Likewise, a 
major research direction is in the use of structured output 
representations, which have been applied in a broad range of areas 
including several of the foregoing examples (for instance, the output 
required of the learning algorithm may be a probabilistic model, a 
graph, or a ranking).  In particular, kernel methods have been 
especially fertile in giving rise to efficient and powerful algorithms 
for both structured inputs and outputs, since (as with SVMs) use of the 
"kernel trick" can make the required optimisations tractable: examples 
include large margin Markov networks, graph kernels, and kernels on 
automata.  More generally, kernels between probability measures have 
been proposed (with no a-priori assumptions as to the dependence 
structure), which have motivated particular kernels between images and 
strings.

In NIPS 2004, two workshops took place addressing learning approaches on 
structured domains: Learning on Structured Outputs 
(Bakir,Gretton,Hoffman,Schoelkopf) and Graphical Models and Kernels 
(Smola, Taskar, Vishwanathan). In view of significant and continued 
advances in the field, the present workshop addresses the same area as 
these earlier workshops: to provide an overview of recent theoretical 
and algorithmic foundations for kernels on structured domains, to 
investigate applications that build on these fundamentals, and to 
explore new research directions for future work. The workshop is also 
intended as one element of the Pascal thematic program on learning with 
complex and structured outputs.

Workshop format
%%%%%%%%%%%%%%%

The workshop will last one day, and will include invited talks (of 30 
minutes' duration), submitted talks (15 minutes), and periods of 
moderated and open discussion (20 minutes).  The final discussion will 
provide a  wrap-up session which will summarise the issues raised, so 
that all participants leave the workshop with a clear view of the future 
challenges and open questions which need to be addressed.

Invited speakers
%%%%%%%%%%%%%%%%

Ben Taskar
Jean-Philippe Vert
Matthias Hein

Call for papers
%%%%%%%%%%%%%%%

We invite submissions for 15 minute contributed talks (of which a 
maximum of eight will be accepted). The intended emphasis is on recent 
innovations, work in progress, and promising directions or problems for 
new research. Proposed topics include:

  * Learning when the inputs/outputs are structures

  * Learning from data embedded in structure

  * Graphical models and information geometry

  * Kernels on probability measures

Our focus will be on using kernel methods to deal efficiently with 
structured data. We will also consider work falling outside these 
specific topics, but within the workshop subject area.

If you would like to submit to this session, please send an abstract to 
Arthur Gretton (arthur at tuebingen dot mpg dot de) before October 21. 
Please do not send posters or long documents. Decisions as to which 
proposals are accepted will be sent out on November 05.

Workshop Chairs
%%%%%%%%%%%%%%%

Arthur Gretton (MPI for Biological Cybernetics)
Gert Lanckriet (UC San Diego)
Juho Rousu     (University of Helsinki)
Craig Saunders (University of Southampton)