Call for Papers
Annals of Operations Research
special issue on
Cross-entropy Algorithms for Combinatorial Optimisation,
Rare-event Simulation and Neural Computation
Guest editors:
Pieter-Tjerk de Boer (ptdeboer@cs.utwente.nl)
Dirk P. Kroese (kroese@maths.uq.edu.au)
Shie Mannor (shie@techunix.technion.ac.il)
Reuven Rubinstein (ierrr01@ie.technion.ac.il)
Submission Deadline: December 31, 2002
The Annals of Operations Research (AOR) invites authors to submit
papers to the special issue on Cross-entropy Algorithms for
Combinatorial Optimisation, Rare-event Simulation, and Neural
Computation.
Background
Introduced in 1997 as an adaptive approach to the efficient estimation
of rare-event probabilities, the cross-entropy (CE) method has rapidly
developed into a powerful and versatile technique for both rare-event
simulation and combinatorial optimisation. The method derives its name
from the cross-entropy (or Kullback-Leibler) distance - a well known
measure of "information", which has been successfully employed in
diverse fields of engineering and science, and in particular in neural
computation, for about half a century. The CE method is an iterative
method, which involves the following two phases:
1. Generation of a sample of random data (trajectories, vectors,
etc.) according to a specified random mechanism.
2. Updating the parameters of the random mechanism, on the basis of
the data, in order to produce a "better" sample in the next
iteration.
The significance of the cross-entropy concept is that it defines a
precise mathematical framework for deriving fast, and in some sense
"optimal" updating/learning rules.
The CE method has been successfully applied to a number of difficult
combinatorial optimization problems, including the maximal cut
problem, the traveling salesman problem (TSP), the quadratic
assignment problem, different types of scheduling problems, the clique
problem, the buffer allocation problem (BAP) for production lines, and
combinatorial optimization problems associated with the genome. It is
important to note that the CE method deals successfully with both
deterministic problems, such as the TSP, and noisy (i.e.,
simulation-based) problems, such as the BAP.
In the field of rare-event simulation, CE is used in conjunction with
the importance sampling (IS) technique: the CE method iteratively
optimizes parameters of the distributions used in the simulation.
Applications include the efficient estimation of performance measures
in telecommunication networks and reliability systems.
For more information on the CE method we refer to the web site:
http://www.cs.utwente.nl/~ptdeboer/ce/.
Topics
We welcome the submission of cross-entropy-based papers on all three
topics: combinatorial optimization, rare-event simulation, and neural
computation. In this issue we plan to include several papers on the
application of the CE method for solving classification, clustering,
vector quantization, image reconstruction and Markovian decision
problems under uncertainty, which are associated with the field of
neural computation and learning. We also encourage submissions on the
comparison of the CE method to related techniques such as Ant Colony
Optimisation, Stochastic Steepest Ascent, the EM algorithm and
Simulated Annealing. Finally, theoretical advances to and analytical
insights into the CE method will be of great value. It is our belief
that this volume will stimulate the interest on the theory and
applications of learning and in particular on learning with CE.
Submission
The deadline for submission is December 31, 2002. All papers should be
submitted to Reuven Rubinstein by e-mail as postscript or pdf files.
One of the four editors will be assigned to each submitted paper. The
editor will carefully referee the paper along with two additional
referees. The papers accepted for publication will be delivered to the
Editor-in-Chief, Prof. Peter Hammer by December 31, 2003. The special
issue of AOR is to published sometime during the year 2004.
This archive was generated by hypermail 2b29 : Tue Jun 18 2002 - 09:39:24 PDT