Evolutionary Computation Digest Sunday, May 2, 2010 Volume 24: Issue 5

SUBMISSION ADDRESS:    [log in to unmask]

(UN)SUBSCRIPTION INSTRUCTIONS:  at the bottom of this email


Today's Topics:

 - Evolutionary Kernel Machines - call for book chapters

 - Second call for Papers - Special session on Multiobjective Opt. at META-2010
 - GECCO-2010 Competitions Announcement
 - ICML 2010 Call for Participation


CALENDAR OF GA-RELATED ACTIVITIES: (with EC-Digest issue reference)

NICSO10,IV Intl Wksp on Nature Inspired ..., Granada, Spain,May 12-14 '10(v23n7)
BIOMA'10,4th Intl Conf on Bioinspired ..., Ljubljana, Slovenia, May 20-21(v24n2)
IIS'10,18th Intl Conf on Intel Inform Syst, Siedlce, Poland, June 8-10'10(v24n1)
ICML-10, Intl Conference on Machine Learning, Haifa,Israel,June 21-25'10(v23n10)
GECCO'10,Genetic and Evol Comp Conference, Portland, Oregon,July 7-11 '10(v23n9)
CEC'10,IEEE Congress on Evol. Comp., Barcelona, Spain, July 18-23 2010   (v24n5)
ICES'10,9th Intl Conference on Evolvable Systems, York, UK, Sep 5-8, 2010(v24n2)
ANTS10,7th Intl Conf on Swarm Intelligence, Brussels, Belgium,Sep8-10 '10(v23n9)
PPSN'10,11th Intl Conf on Parallel Problem ..., Krakow, Poland, Sep 11-15(v23n9)
META'10,Intl. Conf. on Meta. and Nature ..., Djerba, Tunisia,Oct28-30'10 (v24n5)
FOGA'11,Foundations of Gen. Alg. XI, Schwarzenberg, Austria, Jan 5-9 2011(v24n3)

Sender: Christian Igel <[log in to unmask]>
Subject: Evolutionary Kernel Machines - call for book chapters

-- Evolutionary Kernel Machines --
Springer, Series: Studies in Computational Intelligence

Oliver Kramer, International Computer Science Institute, Berkeley
[log in to unmask]

Rudolph, Fakultät für Informatik, Technische Universität Dortmund
[log in to unmask]

Christian Igel, Institut für Neuroinformatik, Ruhr-Universität Bochum
[log in to unmask]

Kernel-based techniques are state of the art in machine learning. The various
optimization problems that arise when applying kernel methods are usually solved
using deterministic mathematical programming techniques. However, there are good
reasons to employ stochastic optimization in machine learning because one faces
problems that suffer from numerous local optima, crucial parameter dependencies,
discontinuous objective functions, and/or noise. Stochastic search methods may
help to overcome these problems. In particular, evolutionary computation has
become a rich field for powerful methods in global optimization. They are
embarrassingly parallelizable and thus fairly efficient search methodologies in
distributed computing scenarios. Successful heuristic extensions have been
proposed for special solution space conditions such as multi-objective, multi-
modal, and constrained objective functions. These developments motivate the
application of advanced evolutionary optimizers to kernel machines. Meanwhile,
many results have shown that this is a fruitful union. This edited book invites
papers on all aspects of theory, computation, and application related topics on
the intersection of evolutionary computation and kernel machines. The goal of
the book is not to be another collection of research papers, but to comprise
deep and self-contained chapters representative in their field, and to give a
comprehensive insight into evolutionary kernel machines. The topics may cover
the following fields:

- Evolutionary and stochastic optimization (evolution strategies, particle swarm
  optimization, hybrid meta-heuristics, etc.) in kernel- based machine learning
  (classification, regression, clustering, dimension reduction, manifold
- Evolutionary engines, e.g., embedded optimization problems, for kernel
- Stochastic search for model selection, e.g., for adaptation of kernel- and
- Balancing of conflicting objectives with evolutionary multi- objective
  optimization methods
- Online-tuning and -adaptation of kernel-based machine learning parameters
- Design of experiments and offline-tuning in kernel-based machine learning
- Machine learning in optimization, e.g., meta-modeling with regression
  techniques or clustering as post-analysis
- Parallelization of evolutionary optimizers for kernel-based learning
- Applications of evolutionary kernel-based machine learning algorithms
- Theoretical analyses, e.g., runtime or convergence analyses for evolutionary
  kernel machines

To make sure that your work fits into the project it is recommended to send a
short abstract of your planned contribution in advance.

Submission deadline: October 1, 2010
Notification of the first review: January 15, 2011
Revisions due: March 1, 2011
Final manuscript: April 2011
Publication: Summer 2011

Manuscripts should be prepared according to the Springer style for monographs,
see for more information.  Submissions
should be done via email to the editors or to
[log in to unmask]

Sender: Arnaud Liefooghe <[log in to unmask]>
Subject: Second call for Papers - Special session on Multiobjective Optimization
         at META-2010

Special session on Multiobjective Optimization

as a part of

META 2010
International Conference on Metaheuristics and Nature Inspired Computing
October 28-30, 2010 - Djerba, Tunisia

Important dates
 - Submission (2-page abstract): May 15, 2010
 - Notification of acceptance: July 15, 2010
 - Camera-ready version: September 15, 2010
 - Conference: October 28-30, 2010

The scope of the special session includes (but is not limited to):
 - Adaptation of metaheuristics to multiobjective optimization: local search,
   simulated annealing, tabu search, iterated local search, evolutionary
   algorithms, swarm optimization, scatter search.
 - Algorithmic aspects of metaheuristics for multiobjective optimization: search
   strategies, fitness assignment, diversity preservation, elitism, constraint
   handling, implementation issues, analysis.
 - Hybridization with other techniques: hybrid metaheuristics for multiobjective
   optimization, cooperation with exact approaches, interaction with
   multicriteria decision making.
 - Performance assessment of metaheuristics for multiobjective optimization: new
   performance indicators, theoretical properties and relation between
 - Theoretical aspectsof metaheuristics for multiobjective optimization:
   landscape analysis, problem properties.
 - Application of metaheuristics for multiobjective optimization to academic and
   real-world multiobjective optimization problems: transportation, scheduling,
   timetabling, bioinformatics.

Submission of papers should be in shape of a 2-page abstract (including figures,
tables, and references). Templates are available on the website:
Extended versions of selected papers will be considered for publication in the
following journals: COR (Computers and Operations Research), ITOR (International
Transactions in Operational Research), JMMA (Journal of Mathematical Modelling
and Algorithms).

Program Committee (to be completed)
 - Matthieu Basseur, University of Angers (FR)
 - Clarisse Dhaenens, University of Lille (FR)
 - José R. Figueira, Technical University of Lisbon (PT)
 - Carlos Fonseca, University of Algarve (PT)
 - Laetitia Jourdan, INRIA Lille – Nord Europe (FR)
 - Nicolas Jozefowiez, University of Toulouse (FR)
 - Francisco Luna, University of Malaga (ES)
 - Thibaut Lust, University of Mons (BE)
 - Julian Molina, University of Malaga (ES)
 - Ricardo Takahashi, Federal University of Minas Gerais (BR)

Organizing Committee
 - Arnaud Liefooghe, University of Coimbra (PT)
 - Luis Paquete, University of Coimbra (PT)

Sender: Christian Gagné <[log in to unmask]>
Subject: GECCO-2010 Competitions Announcement

                    GECCO-2010 COMPETITIONS
                    Sponsored by ACM SIGEVO

We are pleased to announce that three competitions will be held in
conjunction with the 2010 Genetic and Evolutionary Computation
Conference (GECCO-2010) in Portland, Oregon, USA, July 7-11.

=-=-=-=-= Evolutionary Art Competition =-=-=-=-=
This competition invites conference participants to demonstrate that genetic and
evolutionary computation can be applied to create impressive and provocative
works of art. The competition will identify the best work, be it an image, a
sculpture, a music score, a video, an interactive online experience, or a system
that exhibits some form of independent creativity.

== Entry Submission ==
Entrants must submit: (1) a brief artistic statement illustrating the concept,
(2) a short paper describing the technical details, and (3) a set of multimedia
files to illustrate the result of the evolutionary process. Artists can either
submit five still images, or a video of up to 5 minutes, or a sound  file of up
to 5 minutes. All submissions should be sent to [log in to unmask] by
June 18, 2010.

== Evaluation ==
The submissions will be evaluated by a jury of researchers from the evolutionary
computation and the technological arts communities, who will evaluate the
submissions on the following criteria: originality (50%), technical quality
(30%), and relevance to evolutionary art and the goals of the competition (20%).

== Presentation ==
Finalists selected by the jury will be invited to present their submission at
the competition session, held during the GECCO conference. The winner of the
competition will be announced at the SIGEVO meeting ceremony, on July 11, 2010.

== Important Dates ==
* Submission deadline: June 18, 2010
* Conference: July 7-11, 2010

== Organizers ==
* Scott Draves, Electric Sheep
* Christian Gagné, Université Laval
* Jeffrey Ventrella, independent artist/researcher

== Calls ==
* PDF:
* swf:

=-=-=-=-=  GPUs for Genetic and Evolutionary Computation =-=-=-=-=
We are pleased to announce the official start of the GPU competition of
GECCO-2010 with the publication of the competition rules and the scoring system.

== The Goal ==
This competition focuses on the applications of genetic and evolutionary
computation that can maximally exploit the parallelism provided by low-cost
consumer graphical cards. The competition will award the best applications both
in terms of degree of parallelism obtained, in terms of overall speed-up, and in
terms of programming style.

== Rules and Regulations ==
Entrants must submit (1) the application sources with the instructions to
compile it and (2) a two page description of the application. Submissions will
be reviewed by a committee of researchers from the evolutionary computation
community and from industry. Each reviewer will score the submission according
to 12 criteria concerning the submitted algorithm, the speed-up it achieves, and
its impact on the evolutionary computation community. The total score will be
obtained as the weighted sum of the 12 separate scores.

Submissions should be mailed to [log in to unmask] no later than
June 23, 2009. The final scores will be announced during GECCO.

== Scoring ==
Submissions will be reviewed by a panel of researchers from the evolutionary
computation community and from industry who will score each submission according
to the following criteria.

* Algorithm (50% of the total score)
 - Novelty (10%): Does the algorithm exploit the GPU in a novel way?
   (e.g., not just for fitness evaluation?)
 - Efficiency (10%): Does the algorithm efficiently use the GPU?
 - GPU-side (10%): How much of the algorithm is implemented GPU side?
 - Elegance (5%): Is the algorithm simple, easy to understand?
 - Portability (5%): Is the code parameterized for different GPU architectures
   and/or across vendors?
 - Suitability (10%): Does it use features of the GPU architecture logically and
   to the advantage of the program?

* Speed (20% of the total score)
 - Speedup (10%): How much is the speed up compared to a well coded CPU version?
 - Resources (5%): What is the resource utilization? (Ideally a program should
   use the 100% of the GPU).
 - Scalability (5%): Will it scale? E.g. to new hardware, multiple GPUs, GPUs
   with fewer/more processors?

* Evolutionary Computation (30% of the total score)
 - Utility (10%): Do the results benefit the EC/GA/GP community?
 - Practicality (10%): Were the results practically obtainable without GPU
 - Science (10%): Is the system used to generate better quality science?

For example, increasing statistical significance, increasing coverage of test
cases or demonstrating greater generalization.

== Important Dates ==
* Submission deadline: June 23th 2010
* Conference: July 7th-11th 2010

== Organizers ==
* Simon Harding, Memorial University of Newfoundland, Canada
* David Luebke, NVIDIA
* Pier Luca Lanzi, Politecnico di Milano
* Edmondo Orlotti, NVIDIA
* Antonino Tumeo, Pacific Northwest National Laboratory, USA

== Calls ==
* PDF:
* swf:

=-=-=-=-=  2010 Simulated Car Racing Championship =-=-=-=-=
We are pleased to announce the start of the 2010 Simulated Car Racing
Championship, an event joining three simulated car racing competitions held at:
* ACM GECCO-2010, Portland (USA), July 7th - 11th
* IEEE WCCI-2010, Barcelona (Spain), July 18th - July 23rd
* IEEE CIG-2010, Copenhagen (Denmark), August 18th - August 21st

== The Goal ==
The goal of the championship is to design a controller for a racing car that
will compete on a set of unknown tracks  first alone (against the clock) and
then against other drivers. The controllers perceive the racing environment
through a number of sensors that describe the relevant features of the car
surroundings, of the car state, and the game state.The controller can perform
the typical driving actions (clutch, changing gear, accelerate, break, steering
the wheel, etc.)

== New Features ==
The 2010 Championship introduces several innovations in comparison with the
2009 edition:
* the range of the proximity sensors have been increased from 100m to 200m
* the position of the range  finders can be customized by users
* new focus sensor provides accurate sensing of the road ahead
* noise is introduced to range and proximity sensors
* drivers now control the clutch
* a new warm-up stage allows drivers to learn about track properties before
  the qualifying

Previous competitors can enter the competition with only small additional effort
since the 2010 APIs are very similar to the ones used for the 2009 edition.

== Rules and Regulations ==
The championship consists of nine races on nine different tracks divided into
three legs, one for each conference, involving three Grand Prix competitions
each. Teams will be allowed to submit a different driver to each leg. Each Grand
Prix consists of three stages: the warm-up, the qualifying, and the race.

During the warm-up, drivers race alone to collect useful information about the
tracks and tune their behaviors. During the qualifying, drivers races alone on
the tracks; the eight fastest drivers participate in the main race. The main
event consists of eight three lap races on each of the three tracks. At the end
of each race, the drivers are scored using the F1 system. The driver performing
the fastest lap in the race will get two additional points. The driver
completing the race with the smallest amount of damage will also get two extra

As in the previous edition, the tracks used in each leg are unknown to
the competitors.

== Championship Legs ==
Each leg will focus on a specific track type and will involve a diff erent
number of laps:
* American Leg (GECCO-2010), three motor speedways and 50 laps.
* Formula Leg (WCCI-2010), three technical (F1-like) tracks and 15 laps.
* Dusty Leg (CIG-2010), three dirt tracks (non-asphalt stretches and bumps)
  and 25 laps.

== Important Dates ==
* American Leg (GECCO-2010):
 - Submission deadline June 27th 2010
 - Conference: July 7th-11th 2010
* Formula Leg (WCCI-2010):
 - Submission deadline: July 7th 2010
 - Conference: July 18th-23rd 2010
* Dusty Leg (CIG-2010):
 - Submission deadline: August 8th 2010
 - Conference: August 18th-21st 2010

== Competition Software ==
The competition software, including servers for Linux & Windows, and C++
and Java clients, can be downloaded from the competition webpage:

For inquiries send an email to [log in to unmask]

== Organizers ==
* Daniele Loiacono (Politecnico di Milano)
* Luigi Cardamone (Politecnico di Milano)
* Martin V. Butz (University of Würzburg)
* Pier Luca Lanzi (Politecnico di Milano)

== Calls ==
* PDF:
* swf:

GECCO is sponsored by the Association for Computing Machinery Special
Interest Group for Genetic and Evolutionary Computation (ACM SIGEVO).

Sender: Johannes Fuernkranz <[log in to unmask]>
Subject: ICML 2010 Call for Participation

                ICML 2010 Call for Participation

The 27th International Conference on Machine Learning (ICML), the premier
conference in the field, will be held in Haifa, Israel, June 21-25, 2010, and
will be co-located with the 23rd Conference on Learning Theory (COLT).

The scientific program features 152 paper and poster presentations, 8 tutorials,
10 workshops, an invited application track, 3 invited speakers, and a guest
lecture by Nobel laureate Robert Aumann.

The social program features a banquet, two evening poster sessions, guided
walking tours through Haifa, and a trip to Jerusalem.

 * Nir Friedman
 * Tom Mitchell
 * Duncan Watts
 * Robert Aumann (Nobel Laureate 2005)

 From 594 submissions, we selected 152 papers, which will be presented in five
 parallel tracks. In addition, each paper will be presented at a poster session.
 We will also have a short track of invited application papers.

 * Domain Adaptation (Hal Daume & John Blitzer)
 * Geometric Tools for Identifying Structure in Large Social and Information
   Networks (Michael W. Mahoney)
 * Kernel Adaptive Filters (José Príncipe & Weifeng Liu)
 * Learning through Exploration (Alina Beygelzimer and John Langford)
 * Metric Learning (Brian Kulis)
 * Privacy-preserving Data Mining (Stan Matwin)
 * Sparse Modeling: Theory, Algorithms and Applications (Irina Rish & Genady
 * Stochastic Optimization for Machine Learning (Nathan Srebro and Ambuj Tewari)

 * Budgeted Learning,
   Dragos Margineantu, Russell Greiner, Tomas Singliar and Prem Melville
 * Learning from Multi-Label Data,
   Min-Ling Zhang, Grigorios Tsoumakas and Zhi-Hua Zhou
 * Learning in Non-(geo)metric Spaces,
   J. Buhmann, R. Duin, M. Figueiredo, E. Hancock, V. Murino and M. Pelillo
 * Learning to Rank Challenge,
   Olivier Chapelle and Yi Chang
 * Machine Learning and Games,
   Kurt Driessens, Olana Missura and Christian Thurau
 * Machine Learning and Open Source Software,
   Soeren Sonnenburg, Mikio Braun, Cheng Soon Ong and Patrik Hoyer
 * Reinforcement Learning and Search in Very Large Spaces,
   Peter Auer, Samuel Kaski and Csaba Szepesvari
 * Social analytics: Learning from Human Interactions,
   Elad Yom-Tov, Shie Mannor and Yossi Richter
 * Topic Models; Structure, Applications, Evaluation, and Extensions,
   Michal Rosen-Zvi, Amit Gruber and Richard Zemel
 * Web Applications of Explore/Exploit Schemes,
   Ronny Lempel, Deepak Agarwa, Yehuda Koren and Bee-Chung Chen


 The early registration deadline is on MAY 9th.

Don't hesitate to contact us at [log in to unmask] , if you have
any questions.

We hope to see you Haifa!

Stefan Wrobel (General Chair)
Thorsten Joachims, Johannes F ürnkranz (Program Co-chairs)
Shai Fine, Dan Pelleg (Local Chairs)


 - Send submissions (articles) to [log in to unmask]
 DO NOT send submissions to the [log in to unmask] address.

 - To subscribe send email to [log in to unmask] containing the
 following text in the body of the message:

     subscribe ec-digest-l <Your Name (up to 4 words)>

 - To unsubscribe send email to [log in to unmask] containing the
 following text in the body of the message:

     unsubscribe ec-digest-l

 - To change your email address, simply unsubscribe the old address
 and subscribe the new one.

 - Send other administrative requests to [log in to unmask]

 - For further information about the digest, see


 End of Evolutionary Computation Digest