Print

Print


================
CEC-BBOB: BLACK-BOX OPTIMIZATION BENCHMARKING at IEEE CEC'2015 (CEC-BBOB)

Special session on Unbounded Real-Parameter Blackbox Optimization

organized as part of the

IEEE Congress on Evolutionary Computation, May 25-28, 2015, Sendai, Japan
================

Benchmarking of optimization algorithms is crucial to assess performance 
of optimizers quantitatively, understand weaknesses and strengths of each 
algorithm and is the compulsory path to test new algorithm designs. 

A thorough benchmarking methodology has been defined in [1] and is 
implemented within the COCO framework (http://coco.gforge.inria.fr/) 
that furnishes most of the tedious tasks of benchmarking 
for the participants:
* choice of well-motivated single-objective benchmark functions 
  and their implementation in Matlab, C, Java, R, and Python,
* design of an experimental set-up,
* generation of data output, and
* post-processing and presentation of the results in graphs and tables 
  (up to already prepared LaTeX templates for writing papers).

Note that for the CEC-BBOB-2015 special session, we provide essentially 
the same test-suite as in the previous editions of BBOB held at GECCO. 
We provide 2 testbeds,
* noise-free and 
* noisy,
and we distinguish between 
* an expensive optimization scenario (where a focus on the first 
  100D function evaluations is assumed) and 
* a general scenario for which we do not limit the maximal number 
  of function evaluations made. 
The participants can freely choose any or all of them. We entirely 
ban different parameter settings for different test functions and 
encourage analyses that study the impact of parameter setting changes.

[1] Real-Parameter Black-Box Optimization Benchmarking: Experimental Setup, 
http://coco.lri.fr/downloads/download13.09/bbobdocexperiment.pdf.


The Special Session Papers
==========================
We encourage any submission that is concerned with black-box optimization 
benchmarking of continuous optimizers, for example papers that:
* decribe and benchmark new or not-so-new algorithms on the CEC-BBOB-2015 
  testbed,
* compare new or existing algorithms from the COCO database, or 
* analyze the data obtained in previous editions of BBOB. 

Similarly to the BBOB-2013 edition, we also focus on benchmarking 
optimization algorithms for expensive optimization (with limited budget) 
which especially invites to benchmark surrogate-assisted algorithms 
(e.g. based on kriging, support vector machines etc.). Participants are 
also encouraged to use the existing database for statistical analyses or 
for designing a portfolio of algorithms.

This session is related to the special session and competition 
on bound constrained optimization organized by 
Ponnuthurai Nagaratnam Suganthan et al. 
(http://www.ntu.edu.sg/home/EPNSugan/index_files/CEC2015/CEC2015.htm). 
Submissions to both sessions are encouraged while we require here 
that papers presenting benchmarking results follow the benchmarking 
methodology of the COCO framework. Papers discussing benchmarking 
methodologies in general are also welcome.


Organisation of the Special Session during the CEC conference
=============================================================
During the special session, algorithms and results will be presented 
by the participants. An overall analysis and comparison will be accomplished

by the organizers and all submitted papers will be critically reviewed 
as for any other CEC'2015 paper. A planned plenary discussion on future 
improvements will, among others, address the question, of how the testbed 
should evolve.


Support Material and Downloads
==============================
All the support material needed to benchmark your algorithm(s), process the
results, and compile the final papers will be available at the CEC-BBOB-2015

page (http://coco.gforge.inria.fr/doku.php?id=cec-bbob-2015).

The final code and LaTeX templates following the CEC'2015 guidelines 
will be available in early November. To ensure/check that your optimizer 
and the COCO framework work well together, the currently available downloads

(for past BBOB workshops) can be used. The code for the final experiments 
and post-processing obey the very same interface. Subscribe for 
special session announcements (see below) to be notified on updates.


Important dates
===============
    01/11/2014: code released
    19/12/2014: submission deadline
    20/02/2014: acceptance notification
    13/03/2015: final paper version due
    25-28/05/2015: workshop at CEC'2015


Contact and Mailing List
========================
Subscribe to our discussion mailing list by following this link 
http://lists.lri.fr/cgi-bin/mailman/listinfo/bbob-discuss.

To receive announcement about the workshop, send an email to the BBOB team 
at bbob_at_lri.fr with title "register to BBOB announcement list".


Organization Committee
======================
Youhei Akimoto, Shinshu University, Nagano, Japan
Anne Auger, Inria Saclay - Ile-de-France, Orsay, France
Dimo Brockhoff, Inria Lille - Nord Europe, Villeneuve d'Ascq, France
Nikolaus Hansen, Inria Saclay - Ile-de-France, Orsay, France
Olaf Mersmann, TU Dortmund University, Dortmund, Germany
Petr Pošík, Czech Technical University, Prague, Czech Republic