conference program and abstracts - tu kaiserslauternconference venue fraunhofer center...
TRANSCRIPT
7TH. INTERNATIONAL CONFERENCE ON HYBRID INTELLIGENT SYSTEMS
17-19 September 2007 Fraunhofer-Zentrum, Kaiserslautern, Germany
Organized by Institute of Integrated Sensors Systems (ISE), TU Kaiserslautern
http://his07.hybridsystem.com/
Conference Program and Abstracts General Co-Chairs Andreas König, TU Kaiserslautern, Germany Mario Köppen, Kyushu Institute of Technology, Japan Program Co-Chairs Nikola Kasabov, KEDRI, AUT, New Zealand Ajith Abraham, Norwegian University of Science and Technology, Norway Tutorials/Special Events Chair Christian Igel, Ruhr-University Bochum, Germany Local Organizing Committee S.K. Lakshmanan, P.M. Tawdross, K. Iswandy, TU Kaiserslautern, Germany S. Peters, Fraunhofer ITWM, Germany Technical Co-Sponsors IEEE Systems Man and Cybernetics Society – IEEE SMC Deutsches Forschungsinstitut für künstliche Intelligenz - DFKI Fraunhofer Institut Techno- und Wirtschaftsmathematik - ITWM IEEE Computational Intelligence Society German Chapter – IEEE CIS AK Bildanalyse und Mustererkennung Kaiserslautern - BAMEK The World Federation on Soft Computing – WFSC Phytec Messtechnik GmbH
Welcome from HIS 2007 Organizers
We are very pleased to once again welcome our colleagues to the Seventh International Conference on Hybrid Intelligent Systems (HIS’07) at Kaiserslautern, Germany, during September 17-19, 2007. The HIS series of conferences attracts a wide range of interests with the development of the next generation of intelligent systems and its various practical applications. A fundamental stimulus to the investigation of hybrid intelligent systems is the awareness in the academic communities that combined approaches will be necessary if the real challenges in artificial intelligence are to be solved. Current research interests in this field focus on integration of the different computing paradigms such as fuzzy logic, neuro-computation, evolutionary computation, probabilistic computing, intelligent agents, machine learning, and other intelligent computing frameworks. There is also a growing interest in the role of sensors, their integration and evaluation in such frameworks. The phenomenal growth of hybrid intelligent systems and related topics has created the need for this international conference as a venue to present the latest research. HIS’07 builds on the success of last year’s. HIS’06, which was held in conjunction with the Fourth Conference on Neuro-Computing and Evolving Intelligence (NCEI 06') in Auckland, New Zealand, December 13-15, 2006, attracted participants from 15 countries. As evident from the general philosophy of HIS conferences, we have a focus on interdisciplinary approaches and the global strategy to bring together research scientists from the various disciplines related to hybrid intelligent systems. HIS’07, the Seventh International Conference on Hybrid Intelligent Systems addresses the following important themes:
• Novel & hybrid methods of soft computing, neural & evolutionary computing, signal & image processing
• Methodology & frameworks for automated HIS design • HIS applications • Evolving/adaptive HIS • Intelligent Sensor systems • Self-monitoring sensor systems • Organic Computing systems • Wireless intelligent sensor systems • Distributed intelligent systems/sensor networks • Bio-inspired HIS hardware
HIS’07 is technically co-sponsored by IEEE Systems Man and Cybernetics Society, Deutsches Forschungsinstitut für künstliche Intelligenz (DFKI), Fraunhofer Institut Techno- und Wirtschaftsmathematik (ITWM), the German Chapter of IEEE Computational Intelligence Society, the Arbeitskreis Bildanalyse und Mustererkennung Kaiserslautern (BAMEK), The World Federation on Soft Computing (WFSC), and Phytec Messtechnik GmbH.
The HIS’07 program committee represented 23 countries on 5 continents. All submitted papers were refereed by at least three independent referees and, in some uncertain cases the number of referees was up to four. Based on the referee’s reports, the Program Committee selected 52 papers for oral presentation and 13 papers for poster presentation. We would like to thank the HIS’07 international program committee and the additional reviewers for providing the reviews in time.
HIS'07 also received four special session proposals, "Ambient Intelligence [at Home and Recreation]" proposed by the Priority Research Center on Ambient Intelligence at TU Kaiserslautern, "Interdisciplinary/Hybrid Approaches to Dependable and Flexible Networking" proposed by the Kyushu Institute of Technology and KDDI R&D Labs, Japan, "Embedded Neural Network Hardware" proposed by the ITG working group 8.4.9. Microelectronics for Neural Networks, and "Aspects of Image Processing: Theory, applications & computations" proposed by Fraunhofer ITWM. Four excellent tutorials will also be presented at the venue. Our special thanks also go to all the plenary speakers for providing the very interesting and informed talks to catalyze subsequent discussions. Our special thanks to Ms. Lisa O'Conner of IEEE Computer Society Press for all the support and help related to the production of this important scientific work. Finally, we would like to express our sincere gratitude to all the authors and local organizing committees that have contributed towards the success of this conference. We look forward to seeing you in Kaiserslautern, Germany, during HIS’07. General Co-chairs Andreas König University of Kaiserslautern, Germany
Mario Köppen Kyushu Institute of Technology, Japan Program Chairs Nikola Kasabov KEDRI, Auckland University, New Zealand
Ajith Abraham Norwegian University of Science and Technology, Norway
International Program Committee
Janos Abonyi, University of Veszprem, Hungary Bruno Apolloni, Universita degli Studi di Milano, Italy Akira Asano, Hiroshima University, Japan Thomas Breuel, U. Kaiserslautern and DFKI, Germany Will Browne, University of Reading, UK Oscar Castillo, Tijuana Institute of Technology, Mexico Sung-Bae Cho, Yonsei University, Korea Leandro Coelho, Pontifical Catholic University of Parana, Brazil David Corne, Heriot-Watt University, UK Bernard De Baets, Ghent University, Belgium Suash Deb, National Institute of Science & Technology, India Marco Degemmis, University of Bari, Italy Joachim Diederich, American University of Sharjah, UAE Richard Duro, University of Coruna, Spain Achim Ebert, University of Kaiserslautern, Germany Katrin Franke, Norwegian Information Security lab (NISlab), Norway Maria Ganzha, EUH-E, Poland Tom Gedeon, Australian National University, Australia Bernard Grabot, LGP-ENIT, France Crina Grosan, University Babes Boyai, Romania Jerzy W Grzymala-Busse, University of Kansas, USA Tatiana Valentine Guy, Acad. of Sciences of the Czech Republic, CZ Hans Hagen, University of Kaiserslautern, Germany Saman Halgamuge, University of Melbourne, Australia Ilkka Havukkala, Auckland University of Technology, New Zealand Francisco Herrera, University of Granada, Spain Frank Hoffmann, University of Dortmund, Germany Christian Igel, Ruhr-University Bochum, Germany Silviu Ionita, University of Pitesti, Romania Hisao Ishibuchi, Osaka Prefecture University, Japan Yoshiteru Ishida, Toyohashi University of Technology, Japan Janina Jakubczyc, Wroclaw University of Economics, Poland Yaochu Jin, Honda Research Institute Europe, Germany Miroslav Karny, Academy of Sciences of Czech Republic, CZ Etienne Kerre, Ghent University, Belgium Frank Klawonn, University of Applied Sciences BS/WF, Germany Joshua Knowles, The University of Manchester, UK Galina Korotkikh, Central Queensland University, Australia Victor Korotkikh, Central Queensland University, Australia Pasquale Lops, Universita' degli Studi di Bari, Italy Sebastian Lozano, University of Seville, Spain Stephen MacDonell, Auckland University of Technology, NZ Corrado Mencar, University of Bari, Italy Michael Defoin Platel, Auckland University of Technology, NZ Takashi Morie, Kyushu Institute of Technology, Japan Luiza de Macedo Mourelle, State University of Rio de Janeiro, Brasil Kazumi Nakamatsu, University of Hyogo, JAPAN Nadia Nedjah, State University of Rio de Janeiro, Brazil Maria do Carmo Nicoletti, Univ. Federal de S. Carlos - SP, Brazil Ikuko Nishikawa, Ritsumeikan University, Japan Hajime, Nobuhara, University of Tsukuba, Japan Jae Oh, Syracuse University, USA Nikhil Pal, Indian Statistical Institute, India Vasile Palade, Oxford University, UK Marcin Paprzycki, SWPS and IBS PAN, Poland Witold Pedrycz, University of Alberta, Canada Francisco Pereira, Universidade de Coimbra, Portugal Radu-Emil Precup, Politehnica University of Timisoara, Romania Javier Ruiz-del-Solar, Universidad de Chile, Chile Ronald Rösch, Fraunhofer ITWM, Germany Ashraf Saad, Armstrong Atlantic State University, USA Karl Sammut, Flinders University, Australia Udo Seiffert, IPK Gatersleben, Germany Giovanni Semeraro, University of Bari, Italy Bernhard Sick, University of Passau, Germany Aureli Soria Frisch, Pompeu Fabra University, Spain Eiji Uchino, Yamaguchi University, Japan Berend Jan van der Zwaag, University of Twente, Netherlands
Thanos Vasilakos, FORTH, Crete, Greece Christian Veenhuis, Fraunhofer IPK, Germany Marley Vellasco, PUC-Rio, BRAZIL Michael N. Vrahatis, University of Patras, Greece Richard Weber, University of Chile, Chile Ronald R. Yager, Iona College, USA Kaori Yoshida, Kyushu Institute of Technology, Japan Yan-Qing Zhang, Georgia State University, USA Uwe Zimmer, The Australian National University, Australia Additional Reviewers Anna Lisa Gentile Antonio Varlaro Dirceu Cavendish Georg Frey Hans-Georg Stark Hiroyoshi Miwa Karla Figueiredo Kei Ohnishi Leo Iaquinta Masaki Aida Masato Tsuru Masato Uchida Masayoshi Kobayashi Omar Paranaiba Oriana Licchelli Pierpaolo Basile Thorsten Suttorp Tobias Glasmachers Verena Heidrich-Meisner
Conference Venue Fraunhofer Center Kaiserslautern Fraunhofer-Platz 1 • 67663 Kaiserslautern (alternative GPS address: Trippstadter Straße 125 ) Getting there By car: Coming from the West on Autobahn A 6, take the exit Kaiserslautern-West (15), then go towards downtown and follow the signs towards the university. Before you get to the university, you will reach the Fraunhofer Center a few hundred meters down Trippstadter Straße, on the right side. Coming from the East on Autobahn A 6 or A 63, take the exit “Autobahndreieck Kaiserslautern” (16a or 15) and follow the sign “Stadtmitte”, then “Universität”. It is best to use the detour behind the train station via Zollamtstraße; at the end of the street, turn onto Trippstadter Straße. By rail and bus: From Kaiserslautern-Hauptbahnhof (main station), you may take a taxi (approx. 2 km) or the bus from stop “Post” (walking distance from main station: appr. 3 min.) no. 6 to “Mölschbach” resp. no. 15 to “Universität”. Get off the bus at “Fraunhofer-Zentrum”. By air: From Frankfurt Rhein-Main-Airport, either by train (approx. 90 min.) or rental car (approx. 60 min.); from Saarbrücken-Airport by rental car (approx. 60 min.)
Car parking: The conference venue offers a limited number of visitors parking site at the front entrance, which are subject to availability 1. The nearest public carpark is at the main station (Hauptbahnhof) with convenient bus/taxi connection to the conference site.
1 Cars parked anywhere off the designated visitors locations will be towed away and retrieved at the owner’s expense.
Spec
ial S
essi
on o
n E
mbe
dded
Neu
ral
Har
dwar
e Z0
3.08
Spec
ial S
essi
on o
n In
terd
isci
plin
ary/
H
ybri
d A
ppro
ache
s to
Dep
enda
ble
and
Flex
ible
Net
wor
king
Z0
3.08
Plen
ary
4, Z
02.0
2 V
isua
l Obj
ect C
lass
Rec
ogni
tion
com
bini
ng
Gen
erat
ive
and
Dis
crim
inat
ive
Met
hods
Pr
of. B
ernt
Sch
iele
Spec
ial S
essi
on
on
Am
bien
t In
telli
genc
e Z0
2.02
Lun
ch B
reak
, Foy
er
Plen
ary
5, Z
02.0
2 E
volv
ing
Con
nect
ioni
st a
nd H
ybri
d Sy
stem
s:
Met
hods
, Too
ls, A
pplic
atio
ns
Prof
. Nik
ola
Kas
abov
Hyb
rid
Info
rmat
ion
Proc
essi
ng II
Z0
2.02
Clo
sing
of t
he C
onfe
renc
e G
ener
al A
nnou
ncem
ents
A
nnou
ncem
ent f
or IC
ON
IP’0
8
Wed
nesd
ay, 1
9 Se
ptem
ber
9.00
a.
m.
to
10.0
0 a.
m.
10.0
0 a.
m.
to
12.0
0 a.
m.
12.0
0 a.
m.
1.00
p.
m. t
o 2.
00
p.m
.
2.00
p.
m. t
o 3.
30
p.m
.
3.30
p.
m.
to
4.00
p.
m.
Hyb
rid
Met
aheu
rist
ics a
nd
Lea
rnin
g II
Z0
3.08
Hyb
rid
Mod
els I
Z0
3.08
Hyb
rid
Sign
al
Proc
essi
ng II
Z0
3.08
Plen
ary
2, Z
02.0
2 A
dvan
ced
Lea
rnin
g of
SO
R N
etw
ork
Em
ploy
ing
Eva
luat
ion-
base
d T
opol
ogy
Rep
rese
ntin
g N
etw
ork
Prof
. Tak
eshi
Yam
akaw
a
Hyb
rid
Sign
al
Proc
essi
ng I
Z02.
02 Cof
fee
Bre
ak, F
oyer
Spec
ial S
essi
on
on A
spec
ts o
f Im
age
Proc
essi
ng
Z02.
02 Lun
ch B
reak
, Foy
er
Plen
ary
3, Z
02.0
2 Pr
efer
ence
Art
icul
atio
n in
Evo
lutio
nary
M
ultio
bjec
tive
Opt
imiz
atio
n Pr
of. C
arlo
s M. F
onse
ca
Hyb
rid
Mod
els I
I Z0
2.02
Cof
fee
Bre
ak, F
oyer
Post
er S
essi
on
Foye
r
Con
fere
nce
Din
ner
at
See
hote
l Gel
ters
woo
g,
Am
Gel
ters
woo
g 20
, Kai
sers
laut
ern
Tue
sday
, 18
Sept
embe
r
8.30
a.m
. to
9.
30 a
.m.
9.30
a.m
. to
10
.30
a.m
.
10.3
0 a.
m.
11.0
0 a.
m.
to
12.3
0 a.
m.
12.3
0 a.
m.
1.30
p.m
. to
2.
30 p
.m.
2.30
p.m
. to
3.
30 p
.m.
3.30
p.m
.
4.00
p.m
. to
5.
30 p
.m.
6.30
p.m
.
Tut
oria
l 2, Z
03.0
8 C
ombi
ning
Gra
dien
t-B
ased
With
Evo
lutio
nary
O
nlin
e L
earn
ing:
An
Intr
oduc
tion
to L
earn
ing
Cla
ssifi
er S
yste
ms
Dr.
Mar
tin V
. But
z
Tut
oria
l 4, Z
03.0
8 C
hem
ical
Dyn
amic
s for
In
telli
gent
Sys
tem
s D
r. Pe
ter D
ittri
ch
Hyb
rid
Info
rmat
ion
Proc
essi
ng I
Z03.
08
Hyb
rid
Met
aheu
rist
ics a
nd
Lea
rnin
g I
Z03.
08
Tut
oria
l 1, Z
02.0
2 A
nt c
olon
y op
timiz
atio
n an
d H
ybri
diza
tions
D
r. C
hris
tian
Blum
Cof
fee
Bre
ak, F
oyer
Tut
oria
l 3, Z
02.0
2 O
rgan
ic C
ompu
ting
for
Vid
eo A
naly
sis
Dr.
Rolf
Wür
tz
Lun
ch B
reak
, Foy
er
Ope
ning
of t
he C
onfe
renc
e, Z
02.0
2
Plen
ary
1, Z
.02.
02
Pare
to-B
ased
Mul
t-O
bjec
tive
Mac
hine
Lea
rnin
g D
r. Ya
ochu
Yin
Hyb
rid
Syst
em D
esig
n I
Z02
.02
Cof
fee
Bre
ak, F
oyer
Hyb
rid
Syst
em D
esig
n II
Z0
2.02
Bus
leav
ing
conf
eren
ce si
te f
or
Wel
com
e R
ecep
tion
at ‘2
1’,
Will
y-B
rand
t-Pla
tz 1
, Kai
sers
laut
ern
HIS
07 P
rogr
am a
t a G
lanc
e
Mon
day,
17
Sept
embe
r
8.30
a.m
. to
10
.00
a.m
.
10.0
0 a.
m.
10
.30
a.m
. to
12
.00
a.m
.
12.0
0 a.
m.
1.00
p.m
.
1.30
p.m
. to
2.
30 p
.m.
2.30
p.m
. to
3.
50 p
.m.
3.50
p.m
.
4.20
p.m
. to
5.
40 p
.m.
6.00
p.m
. to
9.
00 p
.m.
HIS07 Detailed Program Session Date &
Time Room Title
Tutorial 1 17 September
8.30 a.m. to 10.00 a.m.
Z02.02 Ant colony optimization: Introduction and Hybridizations Dr. Christian Blum ALBCOM, Dept. Llenguatges i Sistemes Informátics Universitat Politècnica de Catalunya, Barcelona , Spain
Tutorial 2
17 September
8.30 a.m. to 10.00 a.m.
Z03.08 Combining Gradient-Based With Evolutionary Online Learning: An Introduction to Learning Classifier Systems Dr. Martin V. Butz Department of Cognitive Psychology III University of Würzburg, Würzburg, Germany
10.00 a.m. to 10.30 a.m. Foyer Coffee break
Tutorial 3
17 September
10.30 a.m. to 12.00 a.m.
Z02.02 Organic Computing for Video Analysis Dr. Rolf Würtz Institut für Neuroinformatik Ruhr-Universität Bochum, Bochum, Germany
Tutorial 4
17 September
10.30 a.m. to 12.00 a.m.
Z03.08 Chemical Dynamics for Intelligent Systems Dr. Peter Dittrich Bio Systems Analysis Group Institute of Computer Science Friedrich-Schiller-University Jena, Jena, Germany
12.00 a.m. to 1.00 p.m. Foyer Lunch break
Opening of Conference
Chair: Prof. Nikola Kasabov
17 September
1 p.m. Z02.02 Prof. Hillebrands, Vice President, TU Kaiserslautern
Prof. Ajith Abraham, Norwegian University of Science and Technology Prof. A. König, ISE, TU Kaiserslautern
Plenary 1
Chair: Prof. Andreas König
17 September
1.30 p.m. to 2.30 p.m.
Z02.02 Pareto-based Multi-Objective Machine Learning Dr. Yaochu Jin Honda Research Institute Europe, Germany
Hybrid Systems Design I
Chair: Prof. Takashi Morie
17 September
2.30 p.m. to 3.50 p.m.
Z02.02 HSDI 1: An FPGA-based Collision Warning System Using Hybrid Approach Haichao Liang, Takashi Morie, Youhei Suzuki, Kazuki Nakada, Tsutomu Miki, and Hatsuo Hayashi
HSDI 2: A Novel Hybrid Training Method for Hopfield Neural Networks Applied to Routing in Communications Networks Wesnaida Schuler, Carmelo Bastos-Filho, and Adriano Oliveira
HSDI 3: On the Use of the SVM Approach in Analyzing an Electronic Nose Manlio Gaudioso, Walaa Khalaf, and Calogero Pace
HSDI 4: A Hybrid Content-Collaborative Recommender System Integrated into an Electronic Performance Support System Leo Iaquinta, Anna Lisa Gentile, Pasquale Lops, Marco de Gemmis, and Giovanni Semeraro
Hybrid Information Processing I
Chair: Dr. Yaochu Jin
17 September
2.30 p.m. to 3.50 p.m
Z03.08 HIP 1: Interpretable Granulation of Medical Data with DC* Corrado Mencar, Arianna Consiglio, and Anna Maria Fanelli
HIP 2: A Clustering Method for Mixed Feature-Type Symbolic Data using Adaptive Squared Euclidean Distances Renata Souza and Francisco Carvalho
HIP 3: A Partitioning Fuzzy Clustering Algorithm for Symbolic Interval Data based on Adaptive Mahalanobis Distances Camilo P. Tenório, Francisco T. De Carvalho, and Julio T. Pimentel
HIP 4: A Multi-Objective Genetic Algorithm for Discovering Non-Dominated Motifs in DNA Sequences Mehmet Kaya
Session Date & Time
Room Title
3.50 p.m. to 4.20 p.m. Foyer Coffee break
Hybrid Systems Design II
Chair: Prof. Andre Ponce de Leon F. de Carvalho
17 September
4.20 p.m. to 5.40 p.m.
Z02.02 HSDII 1: Comparing Several Evaluation Functions in the Evolutionary Design of Multiclass Support Vector Machines Ana Carolina Lorena and André Carlos P. L. F. de Carvalho
HSDII 2: Artificial Immune System with ART Memory Hybridization Jose Lima Alexandrino, Cleber Zanchettin, and Edson Costa de Barros Carvalho Filho
HSDII 3: Evaluating the Performance of a Biclustering Algorithm Applied to Collaborative Filtering – A Comparative Analysis Pablo Dalbem de Castro, Fabrício de França, Hamilton Ferreira, and Fernando Von Zuben
HSDII 4: Computing Sharp Lower and Upper Bounds for the Minimum Latency Problem João Sarubbi, Henrique Luna, Gilberto Miranda Jr., and Ricardo Camargo
Hybrid Metaheuristics and Learning
I Chair: Prof. Teresa Ludermir
17 September
4.20 p.m. to 5.40 p.m.
Z03.08 HMLI 1: A Cooperative System of Metaheuristics Jose Manuel Cadenas, Maria del Carmen Garrido, and Enrique Muñoz
HMLI 2: Active Learning to Support the Generation of Meta-Examples Ricardo Prudencio and Teresa Ludermir
HMLI 3: Block Encryption Using Hybrid Additive Cellular Automata Petre Anghelescu, Silviu Ionita, and Emil Sofron
HMLI 4: Symbiotic Tabu Search, A General Evolutionary Optimization Approach Ramin Halavati, Saeed Bagheri Shouraki, Bahareh Jafari Jashmi, and Mojdeh Jalali Heravi (Presentation canceled)
17 September
6.00 p.m. to 9.00 p.m.
Bus leaving conference site for Welcome Reception at 21,
Willy-Brandt-Platz, Kaiserslautern, (www.21-lounge.de), Reception ends at 9.00 p.m., participants are welcome to stay as regular guests
Session Date & Time
Room Title
Plenary 2 Chair: Prof. Andreas König
18 September
8.30 a.m. to 9.30 a.m.
Z02.02 Advanced Learning of SOR Network Employing Evaluation-based Topology Representing Network Prof. Takeshi Yamakawa, Co-authors: Keiichi Horio and Takahiro Tanaka Dep. of Brain Science and Engineering, Graduate School of Life Science and Systems Eng., Kyushu Institute of Technology, Japan
Hybrid Signal Processing I
Chair: Prof. Christian Igel
18 September
9.30 a.m. to 10.30 a.m.
Z02.02 HSPI 1: Image Retrieval Using the Curvature Scale Space (CSS) Technique and the Self-Organizing Map (SOM) Model under Affine Transforms Carlos de Almeida, Renata de Souza, Carlos Rodrigues, and Nicomedes Cavalcanti Júnior
HSPI 2: Particle Detection on Electron Microscopy Micrographs Using Multi-Classifier Systems Lucas M. Oliveira, Raul B. Paradeda, Bruno M. Carvalho, Anne M. P. Canuto, and Marcilio C. P. de Souto
HSPI 3: Evolutionary Optimization of Wavelet Feature Sets for Real-Time Pedestrian Classification Jan Salmen, Thorsten Suttorp, Johann Edelbrunner, and Christian Igel
Hybrid Metaheuristics and Learning
II Chair: Prof. Ajith Abraham
18 September
9.30 a.m. to 10.30 a.m.
Z03.08 HMLII 1: A New PSO Algorithm Incorporating Reproduction Operator for Solving Global Optimization Problems Millie Pant, Thangaraj Radha, and Ajith Abraham
HMLII 2: Hybridized Swarm Metaheuristics for Evolutionary Random Forest Generation Miroslav Bursa, Lenka Lhotska, and Martin Macas
HML II 3: Visualization of Pareto-Sets in Evolutionary Multi-Objective Optimization Mario Köppen and Kaori Yoshida
10.30 a.m. to 11.00 a.m. Foyer Coffee break
Special Session on Aspects of
Image Processing:
Theory, applications & computations
Chair: Dr. Ronald Rösch
18 September
11.00 a.m. to 12.30 a.m.
Z02.02 SIP 1: Global Modes in Kernel Density Estimation: RAST Clustering Oliver Wirjadi and Thomas Breuel
SIP 2: Fiber Orientation Estimation from 3D Image Data: Practical Algorithms, Visualization, and Interpretation Katharina Robb, Oliver Wirjadi, and Katja Schladitz
SIP 3: A Hybrid Texture Analysis System based on Non-Linear & Oriented Kernels, Particle Swarm Optimization and kNN vs. Support Vector Machines Stefanie Peters and Andreas König
Hybrid Models I
Chair: Prof. Carlos Fonseca
18 September
11.00 a.m. to 12.20 a.m.
Z03.08 HMI 1: Application of a Hybrid Classifier to the Recognition of Petrochemical Odors Eleonora Oliveira, Paulemir Campos, and Teresa Ludermir
HMI 2: An Immune Genetic Algorithm for Software Test Data Generation Abdel Hamid Bouchachia
HMI 3: Fuzzy Neural Network Model Using a Fuzzy Learning Vector Quantization with the Relative Distance Yong Soo Kim and Sung-ihl Kim
HMI 4: Evolutionary Approaches to Solve an Integrated Lot Scheduling Problem in the Soft Drink Industry Claudio Toledo, Paulo França, Reinaldo Morabito, and Alf Kimms
Session Date & Time
Room Title
12.30 a.m. to 1.30 p.m. Foyer Lunch break
Plenary 3 Chair: Dr. Mario Köppen
18 September
1.30 p.m. to 2.30 p.m.
Z02.02 Preference Articulation in Evolutionary Multiobjective Optimisation Prof. Carlos M. Fonseca Centre for Intelligent Systems Faculty of Science and Technology Universidade do Algarve Faro, Portugal
Hybrid Models II
Chair: Dr. Kei Ohnishi
18 September
2.30 p.m. to 3.30 p.m.
Z02.02 HMII 1: Bayesian Networks Modeling for Asian Suprema Soybean Rust Incidence Study in Different Conditions of Temperatures and Leaf Wetness Ricardo Martins de Abreu Silva, Felipe L. Valentim, and Marcelo C. Alves
HMII 2: Hybrid Approach to Solve a Crew Scheduling Problem: an Exact Column Generation Algorithm Improved by Metaheuristics André Santos and Geraldo Mateus
HMII 3: Model and Algorithms for the Multicommodity Traveling Salesman Problem João Sarubbi, Geraldo Mateus, Luna Henrique, and Miranda Jr. Gilberto
Hybrid Signal Processing II
Chair: Prof. Katrin Franke
18 September
2.30 p.m. to 3.30 p.m.
Z03.08 HSPII 1: Content Based Image Retrieval using a Descriptors Hierarchy Raquel Esperanza Patiño Escarcina and Jose Alfredo Ferreira Costa HSPII 2: Pareto-dominated Hypervolume Measure: An Alternative Approach to Color Morphology Mario Köppen and Katrin Franke HSPII 3: A Novel Fully Evolved Kernel Method for Feature Computation from Multisensor Signal Using Evolutionary Algorithms Kuncup Iswandy and Andreas König
3.30 p.m. to 4.00 p.m. Foyer Coffee break
Poster Session
18 September
4.00 p.m. to 5.30 p.m.
Foyer P 1: Genetic Programming meets Model-Driven Development Thomas Weise, Michael Zapf, Mohammad Ullah Khan, and Kurt Geihs
P 2: Particle Swarm Optimization of Neural Network Architectures and Weights Marcio Carvalho and Teresa Ludermir
P 3: Generating Fuzzy Rules from Examples Using the Particle Swarm Optimization Ahmed Esmin
P 4: Dataflow Orchestration of Image Processing Algorithms Using High Level Petri Nets Björn Wagner, Andreas Dinges, and Paul Müller
P 5: Learning Context-Free Grammars from Partially Structured Examples: Juxtaposition of GCS with TBL Olgierd Unold and Lukasz Cielecki - continued next page -
Session Date & Time
Room Title
Poster Session
(continued) Chair:
18 September
4.00 p.m. to 5.30 p.m.
Foyer P 6: Software Effort Estimation using Machine Learning Techniques with Robust Confidence Intervals Petronio Braga, Adriano Oliveira, and Silvio Meira
P 7: Fuzzy and Neuro-Fuzzy Modeling for Total Volume study of Eucalyptus sp Ricardo Martins de Abreu Silva, Adriano R. de Mendonça, Fillipe G. Brandão, Danilo M. Pires, and Gleimar B. Baleeiro
P 8: Dynamic Overlay Networks for Image Processing Grids Andreas Dinges, Bjoern Wagner, and Paul Mueller
P 9: Exploration of Pareto Frontier Using a Fuzzy Controlled Hybrid Line Search Crina Grosan and Ajith Abraham
P 10: Comparative Study of Clustering Techniques for the Organization of Software Repositories Ronaldo Veras, Adriano Oliveira, Bruno Melo, and Silvio Meira
E 1: Exchanging Knowledge and Experience Regarding Identification Systems – Bridging Different Worlds Jochen Müller and Paul Müller
18 September
6.00 p.m. Bus leaving for conference dinner
18 September
6.30 p.m.
Conference Dinner at Seehotel Gelterswoog, Am Gelterswoog 20,
Dinner ends at 11.00 p.m., participants are welcome to stay as regular guests
18 September
23.00 p.m. Bus leaving for hotels
Session Date & Time
Room Title
Plenary 4 Chair: Prof. Ajith Abraham
19 September
9.00 a.m. to 10.00 a.m.
Z02.02 Visual Object Class Recognition combining Generative and Discriminative Methods Prof. Bernt Schiele Multimodal Interactive Systems Department of Computer Science TU Darmstadt, Germany
Special Session on Ambient
Intelligence [at Home and
Recreation]
Chair: Dr. Bernd Schürmann
19 September
10.00 a.m. to 12.00 a.m.
Z02.02 AMI 1: Sensor-based Training Optimization of a Cyclist Group Ankang Le, Thomas Jaitner, and Lothar Litz
AMI 2: Indoor Localisation of Humans, Objects, and Mobile Robots with RFID Infrastructure Jan Koch, Jens Wettach, Eduard Bloch, and Karsten Berns
AMI 3: MacZ - A Quality-of-Service MAC Layer for Ad-hoc Networks Philipp Becker, Reinhard Gotzhein, and Thomas Kuhn
AMI 4: A Hybrid Approach to Intelligent Living Assistance Markus Nick and Martin Becker
Special Session on Embedded
Neural Network
Hardware Chair: Prof. Ulrich Rückert
19 September
10.00 a.m. to 12.00 a.m.
Z03.08 ENNH 1: Neighborhood Rank Order Coding for Robust Texture Analysis and Feature Extraction Christian Mayr and Rene Schüffny
ENNH 2: A Neuromorphic aVLSI Network Chip With Configurable Plastic Synapses Patrick Camilleri, Massimiliano Giulioni, Vittorio Dante, Davide Badoni, and Giacomo Indiveri
ENNH 3: A Digital Framework for Pulse Coded Neural Network Hardware with Bit-Serial Operation Tim Kaulmann, Deniz Dikmen, and Ulrich Rueckert
ENNH 4: Hybrid Intelligent and Adaptive Sensor Systems with Improved Noise Invulnerability by Dynamically Reconfigurable Matched Sensor Electronics Senthil Kumar Lakshmanan and Andreas König
12.00 a.m. to 1.00 p.m. Foyer Lunch break
Plenary 5 Chair: Prof. Kaori Yoshida
19 September
1.00 p.m. to 2.00 p.m.
Z02.02 Evolving Connectionist and Hybrid Systems : Methods, Tools, Applications Prof. Nikola Kasabov Knowledge Engineering and Discovery Research Institute (KEDRI) , Auckland University of Technology, New Zealand
Hybrid Information Processing II
Chair: Dr. Ricardo Silva
19 September
2.00 p.m. to 3.20 p.m.
Z02.02 HIPII 1: How to Obtain Fair Managerial Decisions in Sugar Cane Harvest Using NSGA-II Diogo Pacheco, Tarcísio Lucas, and Fernando Lima Neto
HIPII 2: Markov-Blanket Based Strategy for Translating a Bayesian Classifier into a Reduced Set of Classification Rules Estevam Hruschka Jr, Maria Nicoletti, Vilma Oliveira, and Glaucia Bressan
HIPII 3: Learning to Reach Optimal Equilibrium by Influence of Other Agents Opinion Dennis Barrios Aranibar and Luiz Marcos Garcia Gonçalvez
HIPII 4: Variable Ordering in the Conditional Independence Bayesian Classifier Induction Process: An Evolutionary Approach Estevam Hruschka Jr., Edimilson Santos, and Sebastian Galvao
Session Date & Time
Room Title
Special Session on
Interdiscipli-nary/Hybrid
Approaches to Dependable and Flexible Networking
Chair: Prof. Masato Tsuru
19 September
2.00 p.m. to 3.30 a.m
Z03.08
DFN 1: Performance Trade-off Exploration by Query-Trail-Mediated Topology Reconstruction in Unstructured P2P Networks Kei Ohnishi, Satoshi Nagamatsu, and Yuji Oie
DFN 2: A Control Method of a P2P Network with Small Degree and Diameter Yusuke Sasaki and Hiroyoshi Miwa
DFN 3: Implementation and Experimental Evaluation of On-Line Simulation Server for OSPF-TE Hitomi Tamura, Tsuyoshi Okubo, Yousuke Inoue, Kenji Kawahara, and Yuji Oie
Closing of the Conference
19 September
3.30 p.m. to 4.00 p.m.
Z02.02 Closing Remarks Prof. Nikola Kasabov, KEDRI, Auckland University of Technology Prof. A. König, ISE, TU Kaiserslautern General Announcements Dr. Mario Köppen, Kyushu Institute of Technology, Japan Announcement for ICONIP’08 Prof. Nikola Kasabov, KEDRI, Auckland University of Technology
Plenary 1 Monday, 17. Sept., 1.30 – 2.30. p.m., Z02.02
Pareto-based Multi-Objective Machine Learning
Yaochu Jin Honda Research Institute Europe, Germany
http://www.soft-computing.de/jin.html
Abstract Machine learning is inherently a multi-objective task. Traditionally, however, either only one of the objectives is adopted as the cost function or multiple objectives are aggregated to a scalar cost function. This can mainly attributed to the fact that most conventional learning algorithms can only deal with a scalar cost function. Over the last decade, efforts on solving machine learning problems using the Pareto-based multi-objective optimization methodology have gained increasing impetus, particularly thanks to the great success of multi-objective optimization using evolutionary algorithms and other population-based stochastic search methods. It has been shown that Pareto-based multi-objective learning approaches are more powerful compared to learning algorithms with a scalar cost functions in addressing various topics of machine learning, such as clustering, feature selection, improvement of generalization ability, knowledge extraction, and ensemble generation. This talk provides first a brief overview of Pareto-based multi-objective machine learning techniques. In addition, a number of case studies are provided to illustrate the major benefits of the Pareto-based approach to machine learning, e.g., how to identify interpretable models and models that can generalize on unseen data from the obtained Pareto-optimal solutions. Three approaches to Pareto-based multi-objective ensemble generation are compared and discussed in detail. Most recent results on multi-objective optimization of spiking neural networks will be presented.
Plenary 2 Tuesday, 18. Sept., 8.30 – 9.30. a.m., Z02.02
Advanced Learning of SOR Network Employing Evaluation-based Topology Representing Network
Takeshi Yamakawa, Keiichi Horio and Takahiro Tanaka
Department of Brain Science and Engineering Graduate School of Life Science and Systems Engineering
Kyushu Institute of Technology 2-4 Hibikino, Wakamatsu, Kitakyushu, Fukuoka 808-0196
Japan http://www.brain.kyutech.ac.jp/coe21/index_e.html
Abstract Learning systems such as multi-layer feed-forward neural networks, wavelet networks and so on need appropriate learning data (input data and teaching output data). These methods are not so useful in case when we cannot get the appropriate learning data. Even in this case, it is not so difficult to evaluate the system output for arbitrarily applied input. The learning data of input-output pairs with their evaluations are easily obtained and thus is easily used for modeling the system. SOR (Self-organizing Relationship) network is a modeling tool, which can be established by a set of input-output data and corresponding evaluation. This SOR network can act as a knowledge acquisition system and also act as a fuzzy inference engine. The linkage among the units in competitive layer is fixed and not flexible, and thus not used for complicated systems. In this plenary talk, the advanced learning process is presented for the original SOR network by employing evaluation-based TRN (topology representing network). By this learning, the linkage among the units in the competitive layer can be more flexible and thus used for modeling of much more complicated systems. The application of the SOR network established by this learning process to a manipulation control is also presented.
Plenary 3 Tuesday, 18. Sept., 1.30 – 2.30. p.m., Z02.02
Preference Articulation in Evolutionary Multiobjective Optimisation
Carlos M. Fonseca CSI - Centre for Intelligent Systems Faculty of Science and Technology
Universidade do Algarve Faro Portugal
Abstract
Real-world optimisation problems often involve a number of conflicting criteria, or objectives. Such problems usually admit multiple Pareto-optimal solutions, i.e. solutions, which cannot be improved upon in all objectives simultaneously. In practice, however, acceptable solutions must perform sufficiently well with respect to all objectives, which means that not all Pareto-optimal solutions may be satisfactory.
Evolutionary approaches to multiobjective optimisation have concentrated mainly on the task of approximating the set of Pareto-optimal solutions of a given problem as well as possible, by generating diverse sets of non-dominated alternatives. Subjective information concerning how different combinations of objective values influence the relative quality of a solution is not required, but this approach tends to become impractical as the number of objectives grows. In practice, however, there are many situations in which such preference information is either available a priori or may be acquired during the initial steps of an optimisation run, even if not in a complete form. Incorporating preference information in evolutionary multiobjective optimisation (EMO) algorithms allows the search to concentrate on, and to better approximate, the relevant regions of the Pareto-optimal front.
In this talk, a number of ways in which preference information may be combined with evolutionary search, in order to improve the relevance and the quality of the optimisation results will be discussed, and application examples will be presented. Important aspects of the discussion will include the form in which preference information is initially available, the impact of preference articulation techniques on the optimisation problems to be solved, the quality of the final solutions obtained, and user-related issues, such as visualisation and interaction. The talk will conclude with the identification of some opportunities for future work.
Plenary 4 Wednesday, 19. Sept., 9.00 – 10.00. a.m., Z02.02
Visual Object Class Recognition combining Generative and Discriminative
Methods
Bernt Schiele Multimodal Interactive Systems
Department of Computer Science TU Darmstadt
Germany [email protected]
Abstract
We describe various approaches capable of simultaneous recognition and localization of multiple object classes using a combination of generative and discriminative methods. A first approach uses a novel hierarchical representation allows to represent individual images as well as various objects classes in a single similarity invariant model. The recognition method is based on a codebook representation where appearance clusters built from edge based features are shared among several object classes. A probabilistic model allows for reliable detection of various objects in the same image. A second approach uses a dense representation and a topic distribution model to obtain an intermediate and general representation that is shared across object categories. Combined with discriminative methods these systems show excellent performance on several object categories.
Plenary 5 Wednesday, 19. Sept., 1.00 – 2.00. p.m., Z02.02
Evolving Connectionist and Hybrid Systems: Methods, Tools, Applications
Nikola Kasabov Knowledge Engineering and Discovery Research Institute (KEDRI)
Auckland University of Technology, New Zealand ([email protected]), www.kedri.info
Abstract
Evolving Connectionist Systems (ECOS) are neural network systems that develop their structure, functionality and internal representation through continuous learning from data and interaction with the environment. ECOS can also evolve through generations of populations using evolutionary computation, but the focus of the presentation is on: (1) Adaptive learning and improvement of each individual model; (2) Knowledge representation, knowledge adaptation and knowledge extraction. The learning process can be: on-line, off-line, incremental, supervised, unsupervised, active, sleep/dream, etc.
Principles of different evolving processes from Nature have been used so far to build ECOS. This presentation introduces several models:
• Simple evolving neural networks and evolving neuro-fuzzy systems [1]; • Evolving spiking neural networks, where brain-like spiking neuronal models are used
to incrementally evolve large networks [2]; • Evolving gene interaction networks, that capture dynamic interaction between genes
from time series genetic data [3]; • Evolving neuro-genetic models, where evolving gene networks are integrated with
evolving spiking neural networks to model brain data [3]; • Evolving quantum-inspired neural networks, where quantum principles such as
superposition and entanglement, are used to incrementally select the features and the structure of an evolving connectionist model [2];
• Hybrid models, combining principles and elements from the above [2].
Different ECOS are demonstrated on challenging problems from bioinformatics, medical decision support, adaptive multimodal information processing and biometrics, autonomous robot control, environmental risk prognosis, financial on-line prediction. For some of the demonstrations, a free software environment NeuCom is used (http://www. theneucom.com).
The presentation concludes with some speculations about integrating quantum, genetic and neuronal principles in computational models, along with giving directions for further research.
References:
[1] N.Kasabov, Evolving connectionist systems: Methods and Applications in Bioinformatics, Brain Study and Intelligent Machines, Springer, London, 2002 [2] N.Kasabov, Evolving connectionist systems: The Knowledge Engineering Approach, (second edition) Springer, London, 2007
[3] L.Benuskova and N.Kasabov, Computational Neurogenetic Modelling, Springer, NY, 2007
Keywords: Computational intelligence, Knowledge-based neural networks, Evolving connectionist systems, Gene regulatory networks, Evolutionary computation, Quantum computation, Data mining; Knowledge discovery, Bioinformatics, Brain study, Evolving robots.
Tutorial 1 Monday, 17. Sept., 8.30 – 10.00. a.m., Z02.02
Ant Colony Optimization: Introduction and Hybridizations Christian Blum, Universitat Politècnica de Catalunya
Ant colony optimization (ACO) is an approximate method for tackling combinatorial as well as continuous optimization problems. From an artificial intelligence point of view ACO is a swarm intelligence method, while from the operations research point of view ACO belongs to the class of algorithms known as metaheuristics. The inspiring source of ACO is the foraging behaviour of ant colonies. The indirect communication between the ants via pheromone trails allows them to find shortest paths between their nest and food sources. This functionality of real ant colonies is exploited in ACO algorithms. The aims of this tutorial are twofold. First, we explain the basics of ACO algorithms in order to satisfy participants that have no previous knowledge of this technique. Second, we also deal with more advanced features of ACO with the aim of presenting useful material for participants with some prior knowledge of ACO algorithms. The tutorial starts by introducing the swarm intelligence origins of ACO algorithms. Then, it is shown how to translate the biological inspiration into a technical algorithm. Some of the most successful ACO variants are presented in this context. More advanced features of ACO algorithms are presented in the context of appealing applications. Finally, the tutorial concludes with an interesting recent research topic: the hybridization of ACO algorithms with other optimization techniques.
Tutorial 2 Monday, 17. Sept., 8.30 – 10.00. a.m., Z03.08
Combining Gradient-Based With Evolutionary Online Learning: An Introduction to Learning Classifier Systems
Martin V. Butz, University of Würzburg Learning Classifier Systems (LCSs), introduced by John H. Holland in the 1970s, are rule-based evolutionary online learning systems that combine gradient-based rule evaluation methods with evolutionary-based rule structure learning techniques. Since the introduction of the accuracy-based XCS classifier system by Stewart W. Wilson in 1995, LCSs were shown to be competitive online learning methods applicable to datamining, reinforcement learning, function approximation, and adaptive cognitive systems problems. Hereby, it was shown that performance is machine learning competitive, but learning is taking place online and is often more flexible and highly adaptive. Moreover, problem knowledge can be extracted easily. This tutorial provides a gentle introduction to LCSs and their general functioning. It then focuses on the XCS classifier system. After a short introduction to the system, a theoretical, modular system analysis is put forward that clarifies how and when XCS learns successfully. Finally, successful applications to various problem domains are surveyed. In conclusion, promising future directions of LCS research and application are discussed.
Tutorial 3 Monday, 17. Sept., 10.30 – 12.00. a.m., Z02.02
Organic Computing for Video Analysis Rolf P. Würtz, Ruhr-Universität Bochum
The tutorial will start with a discussion of information technology (IT) problems caused by the rapidly increasing complexity of systems to be deployed. In the world of living beings on the other hand, it can be observed that extremely complex systems function in a robust, fault-tolerant, flexible, adaptive, self-organizing way, and apparently goal-directed way. It is therefore intriguing to identify strategies by means of which these properties are achieved by living systems. This ``Learning from Nature'' is the founding idea of Organic Computing. Earlier IT applications include Neural Networks and Evolutionary Computation. The current interest in Organic Computing is also sustained by a notion of ``Organic'' which relates to the user rather than the developer. In that aspect, Organic Computing requires that the interface between the IT system and the user be organic, intelligible, and friendly. This again imposes constraints on the user interfaces, which can only be partly fulfilled by current technology. The introduction is followed by some facts and theories about self-organizing systems including a short description of current research projects and open issues. One project develops flexible control of traffic for the city of Hanover, which tries to optimizes the overall flow without relying on centralized controllers. Within the automobile, the exploding number of components and interactions and the combinatorics of possible models has prompted the development of an evolutionary architecture which self-organizes according to a goal description and reorganizes in the presence of partial failure. In an ongoing project, the organization of varying office users and people looking for them within the building is handed over to an organic system. The application domain I will present in detail is computer vision and user interaction. First a system for automatic face recognition is described which has been constructed according to neurobiological findings and a theory of self-organizing neural networks. It is also an example for the hierarchical self-organization of elementary feature detectors into structures of higher and higher complexity. Detailed self-organizing neuronal dynamics are presented as well as the techniques of pyramid matching and Elastic Graph Matching, the latter being more efficient on digital computers. The basic matching mechanism is extended to the bunch graph data format and recognition procedure, which has made this technology one of the leading methods for facial identification. This data format can be used to learn facial attributes like ``gender,'' ``beardedness,'' ``wearing glasses'' solely from examples, without need for formulation of rules defining these attributes. In a currently ongoing project, it is used to diagnose certain genetic diseases from facial images. The extension from faces to body gestures is more complicated and partly subject of ongoing research. I present the method of ``democratic integration'', which allows for flexible integration of many fragile cues into a robust decision. This is used for user interaction with a robot gripsee and for interpretation of user gestures. Concluding the tutorial I present a system for the self-organization of a recognition memory for everyday objects. Again the focus is on automatic learning from examples.
Tutorial 4
Monday, 17. Sept., 10.30 – 12.00. a.m., Z03.08
Organization-Oriented Chemical Programming Peter Dittrich and Naoki Matsumaru, Friedrich-Schiller-University Jena
All known life forms process information on a bio-molecular level. This kind of information processing is known to be robust, self-organizing, adaptive, decentralized, asynchronous, fault-tolerant, and evolvable. Computation emerges out of an orchestrated interplay of many decentralized relatively simple components (molecules). Therefore it appears attractive to consider (artificial) chemical information processing as part of hybrid intelligent systems. In this tutorial we will first study fundamental principles of artificial (i.e., in silico) chemical computing and discuss some applications, such as Gamma, MGS, amorphous computing, and reaction-diffusion processors. In the second part we will focus on techniques of ``chemical programming´´. It turns out that in accordance with Conrad's tradeoff principle, programming a chemical computer appears to be difficult and novel techniques are required that help to bridge the micro-macro gap between reaction rules and resulting behavior.
Regular Papers
Session HSDI: Hybrid Systems Design I, Monday, 17. Sept., 2.30 – 3.50. p.m., Z02.02 An FPGA-based Collision Warning System Using Hybrid Approach Haichao Liang, Takashi Morie, Youhei Suzuki, Kazuki Nakada, Tsutomu Miki, Hatsuo Hayashi Abstract: In this paper, we propose an FPGA-based real-time collision warning system for advanced automobile driver assistance systems or autonomous moving robots. The system consists of three function blocks: coarse edge detection using a resistive-fuse network, moving- object detection inspired by neuronal propagation in the hippocampus, and danger evaluation and collision warning using fuzzy inference. The first two functions are implemented in FPGAs. The system can detect moving objects with a speed range of 3-192km/h with a sampling period of 32ms for an input image of 320 x 240 pixels, and can output a warning against dangerous objects. A Novel Hybrid Training Method for Hopfield Neural Networks Applied to Routing in Communications Networks Wesnaida Schuler, Carmelo Bastos-Filho, and Adriano Oliveira Abstract: Efficient routing algorithms are very important for the operation of communication networks, including the Internet. This article proposes a novel hybrid intelligent method for routing which combines Hopfield Neural Networks (HNN) and simulated annealing (SA). The proposed method introduces a modified version of the discrete-time equation used by Bastos-Filho et al [1]. The novel version of the equation aims to improve the HNN convergence, thereby decreasing the computation cost. In our method, the SA algorithm is used to obtain the optimal parameters of the HNN. Simulations reported in this paper shows that the proposed method outperforms the method of Bastos-Filho et al [1], by computing routes using smaller number of iterations and smaller error. On the Use of the SVM Approach in Analyzing an Electronic Nose Manlio Gaudioso, Walaa Khalaf, and Calogero Pace Abstract: We present an Electronic Nose (ENose) which is aimed both at identifying the type of gas and at estimating its concentration. Our system contains 8 sensors, 5 of them being gas sensors (of the class TGS from FIGARO USA, INC., whose sensing element is a tin dioxide (SnO2) semiconductor), the remaining being a temperature sensor (LM35 from National Semiconductor Corporation), a humidity sensor (HIH–3610 from Honeywell), and a pressure sensor (XFAM from Fujikura Ltd.). Our integrated hardware–software system uses some machine learning principles and least square regression principle to identify at first a new gas sample, and then to estimate its concentration, respectively. In particular we adopt a training model using the Support Vector Machine (SVM) approach to teach the system how discriminate among different gases, then we apply another training model using the least square regression, for each type of gas, to predict its concentration. A Hybrid Content-Collaborative Recommender System Integrated into an Electronic Performance Support System Leo Iaquinta, Anna Lisa Gentile, Pasquale Lops, Marco de Gemmis, and Giovanni Semeraro Abstract: An Electronic Performance Support System (EPSS) introduces challenges on contextualized and personalized information delivery. Recommender systems aim at delivering and suggesting relevant information according to users preferences, thus EPSSs could take advantage of the recommendation algorithms that have the
effect of guiding users in a large space of possible options. Collaborative and content-based filtering are the recommendation techniques most widely adopted to date. The main contribution of this paper is a content-collaborative hybrid recommender which computes similarities between users relying on their content-based profiles, in which user preferences are stored, instead of comparing their rating styles. Traditional keyword-based profiles are unable to capture the semantics of user interests, due to the natural language ambiguity. A distinctive feature of our technique is that a statistical model of the user interests is obtained by machine learning techniques integrated with linguistic knowledge contained in WordNet. This model, namely “semantic user profile”, is exploited by the hybrid recommender in the neighborhood formation process. Session HIPI: Hybrid Information Processing I, Monday, 17. Sept., 2.30 – 3.50. p.m., Z03.08 Interpretable Granulation of Medical Data with DC* Corrado Mencar, Arianna Consiglio, and Anna Maria Fanelli Abstract: In this paper we describe an approach for mining interpretable diagnostic rules through a fuzzy information granulation process. Specifically, this process is performed by the DC* algorithm (Double Clustering with A*), which is aimed at mining from data a set of fuzzy information granules that satisfy a number of interpretability constraints. Such granules can be labelled with linguistic terms and used as building blocks for deriving diagnostic rules. The DC* is based on two clustering steps. The first step applies the LVQ1 algorithm to find a number of prototypes in the input space, which represent hidden relationships among data. The second clustering step --based on the A* search-- takes place on the projections of such prototypes, and is aimed at finding an optimal number of granules that verify interpretability constraints. The application of DC* to two well-known medical datasets provided a set of intelligible rules with satisfactory accuracy. A Clustering Method for Mixed Feature-Type Symbolic Data using Adaptive Squared Euclidean Distances Renata Souza and Francisco Carvalho Abstract: This work presents a clustering method for mixed feature-type symbolic data. The presented method needs a previous pre-processing step to transform mixed symbolic data into modal symbolic data. The dynamic clustering algorithm with adaptive distances has then as input a set of vectors of modal symbolic data (weight distributions) and furnishes a partition and a prototype to each class by optimizing an adequacy criterion that measures the fitting between the clusters and their representatives based on adaptive squared Euclidean distances. Examples with synthetic symbolic data sets and an application with a real symbolic data sets show the usefulness of this method. A Partitioning Fuzzy Clustering Algorithm for Symbolic Interval Data based on Adaptive Mahalanobis Distances Camilo P. Tenório, Francisco T. De Carvalho, and Julio T. Pimentel Abstract: The recording of symbolic interval data has become a common practice with the recent advances in database technologies. This paper introduces a fuzzy clustering algorithm to partitioning symbolic interval data. The proposed method furnish a fuzzy partition and a prototype (a vector of intervals) for each cluster by optimizing an adequacy criterion that measures the fitting between the clusters and their representatives. To compare symbolic interval data, the method use a suitable adaptive Mahalanobis disance defined on vectors of intervals. Experiments with real and synthetic symbolic interval data sets showed the usefulness of the proposed method. A Multi-Objective Genetic Algorithm for Discovering Non-Dominated Motifs in DNA Sequences
Mehmet Kaya Abstract: We propose an efficient method using multi-objective genetic algorithm (MOGAMOD) to discover optimal motifs in sequential data. The main advantage of our approach is that a large number of tradeoff (i.e., nondominated) motifs can be obtained by a single run with respect to conflicting objectives: similarity, motif length and support maximization. To the best of our knowledge, this is the first effort in this direction. MOGAMOD can be applied to any data set with a sequential character. Furthermore, it allows any choice of similarity measures for finding motifs. By analyzing the obtained optimal motifs, the decision maker can understand the tradeoff between the objectives. We compare MOGAMOD with the two well-known motif discovery methods, AlignACE, MEME and Weeder. Experimental results on real data set extracted from TRANSFAC database demonstrate that the proposed method exhibits good performance over the other methods in terms of runtime, the number of shaded samples and multiple motifs. Session HSDII: Hybrid Systems Design II, Monday, 17. Sept., 4.20 – 5.40. p.m., Z02.02 Comparing Several Evaluation Functions in the Evolutionary Design of Multiclass Support Vector Machines Ana Carolina Lorena and André Carlos P. L. F. de Carvalho Abstract: Support Vector Machines were originally designed to solve two-class classification problems. When they are applied to multiclass classification problems, the original problem is usually decomposed into multiple binary subproblems. Afterwards, individual classifiers are induced to solve each of these binary problems. To obtain the final multiclass prediction, the outputs of these binary classifiers generated are combined. Genetic Algorithms can be used to optimize the combination of binary classifiers, defining the decomposition according to the performance obtained in the multiclass problem solution. This paper investigates several evaluation functions that can be used in order to evaluate the performance of the decompositions evolved by Genetic Algorithms. Artificial Immune System with ART Memory Hybridization Jose Lima Alexandrino, Cleber Zanchettin, and Edson Costa de Barros Carvalho Filho Abstract: The present work proposes the an architecture Clonart (Clonal Adaptive Resonance Theory) that employs many different techniques like intelligent operatores, clonal selection principle, local search, memory antibodies and ART network in order to increase the performance of the algorithm. The approach uses a mechanism similar to the ART 1 network for storing a population of memory antibodies that will be responsible for the acquired knowledge of the algorithm. This characteristic allows the algorithm a self-organization of the antibodies in accordance with the complexity of the database used. Evaluating the Performance of a Biclustering Algorithm Applied to Collaborative Filtering – A Comparative Analysis Pablo Dalbem de Castro, Fabrício de França, Hamilton Ferreira, and Fernando Von Zuben Abstract: Collaborative filtering (CF) is a method to perform automated suggestions for a user based on the opinion of other users with similar interest. Most of the CF algorithms do not take into account the existent duality between users and items, considering only the similarities between users or only the similarities between items. The authors have proposed in a previous work a bio-inspired methodology for CF, namely BIC-aiNet, capable of clustering rows and columns of a data matrix simultaneously. The usefulness and performance of the methodology are reported in the literature. Now, the authors carry out more rigorous comparative experiments with BIC-aiNet and other techniques found in the literature, as well as evaluate the scalability of the algorithm in several datasets of different sizes. The results indicate that our proposal is able to provide useful recommendations for the users, outperforming other methodologies for CF.
Computing Sharp Lower and Upper Bounds for the Minimum Latency Problem João Sarubbi, Henrique Luna, Gilberto Miranda Jr., and Ricardo Camargo Abstract: The Minimum Latency Problem, also known as Traveling Repairman Problem, the Deliveryman Problem and the Traveling Salesman Problem with Cumulative Costs is a variant of the Traveling Salesman Problem in which a repairman is required to visit customers located on each node of a graph in such a way that the overall waiting times of these customers is mind. In the present work, an algorithm based on tight different linear programming lower bounds and a specialized GRASP provider upper bounds is presented. The linear programming based lower-bounds is base on the Quadratic Assignment Problem and with the help of side constraints. Instances from $10$ up to $60$ nodes are solved to optimality in reasonable time and good upper bounds are also presented. Further ideas on developing specialized algorithms are suggested for future investigations. Session HMLI: Hybrid Metaheuristics and Learning I, Monday, 17. Sept., 4.20–5.40. p.m., Z03.08 A Cooperative System of Metaheuristics Jose Manuel Cadenas, Maria del Carmen Garrido, and Enrique Muñoz Abstract: Hybrid systems give more flexible mechanisms for solving complex problems that can be very difficult to solve using less tolerant approaches. Therefore, a hybrid system will be the most suitable tool in order to cope with the algorithm-instance problem, which says that it is possible that an algorithm and its parameters that obtain good results for an instance of a problem, do not get the same results for another instance of the same problem. All this leads us to use different algorithms to solve combinatorial optimization problems within a single coordinated schema, that is a hybrid cooperative system of metaheuristics. In order to build this system we have proposed a methodology for the construction of a hybrid system, based on Data Mining and Soft Computing. In order to test the usefulness of this methodology two hybrid systems based on a fuzzy model have been constructed to solve the knapsack problem. The first system coordinates two metaheuristics, a Genetic Algorithm and a Tabu Search. The second one adds a third metaheuristic, Simmulated Annealing, in order to check the robustness of the system and its capacity of obtaining higher quality solutions when a metaheuristic is added. Results obtained by this systems and a comparison with the ones obtained with individual metaheuristics are shown. Active Learning to Support the Generation of Meta-Examples Ricardo Prudencio and Teresa Ludermir Abstract: Meta-Learning has been used to select algorithms based on the features of the problems being tackled. Each training example in this context, i.e. each meta-example, stores the features of a given problem and the performance information obtained by the candidate algorithms in the problem. The construction of a set of meta-examples may be costly, since the algorithms performance is usually defined through an empirical evaluation on the problem at hand. In this context, we proposed the use of Active Learning to select only the relevant problems for meta-example generation. Hence, the need for empirical evaluations of the candidate algorithms is reduced. Experiments were performed using the classification uncertainty of the k-NN algorithm as the criteria for active selection of problems. A significant gain in performance was yielded by using the Active Learning method. Block Encryption Using Hybrid Additive Cellular Automata Petre Anghelescu, Silviu Ionita, and Emil Sofron Abstract: With the ever-increasing growth of data communication, the need for security and privacy has become
a strong necessity. In these conditions, the necessity of new powerful encryption techniques becomes a crucial issue. In this paper Cellular Automata (CA) are applied to construct cryptography algorithms. We present an encryption system implemented on a structure of Hybrid Additive Cellular Automata (HACA) used for securing of medical data sent over the internet. The experimental results prove the powerful of cellular automata encryption systems. The method supports both software and hardware implementation. In this paper we present a fully functional software application for data encryption of multimedia medical content. Symbiotic Tabu Search, A General Evolutionary Optimization Approach Ramin Halavati, Saeed Bagheri Shouraki, Bahareh Jafari Jashmi, and Mojdeh Jalali Heravi Abstract: Recombination in the Genetic Algorithm (GA) is supposed to extract the component characteristics from two parents and reassemble them in different combinations – hopefully producing an offspring that has the good characteristics of both parents. Symbiotic Combination is formerly introduced as an alternative for sexual recombination operator to overcome the need of explicit design of recombination operators in GA. This paper presents an optimization algorithm based on using this operator in Tabu Search. The algorithm is benchmarked on two problem sets and is compared with standard genetic algorithm and symbiotic evolutionary adaptation model, showing success rates higher than both cited algorithms. Session HSPI: Hybrid Signal Processing I, Tuesday, 18. Sept., 9.30–10.30. a.m., Z02.02 Image Retrieval Using the Curvature Scale Space (CSS) Technique and the Self-Organizing Map (SOM) Model under Affine Transforms Carlos de Almeida, Renata de Souza, Carlos Rodrigues, and Nicomedes Cavalcanti Júnior Abstract: In a previous work [1], we presented an approach for shape-based image retrieval using the curvature scale space (CSS) and self-organizing map (SOM) methods. Here, we examine the robustness of the representation under affine transforms. Moreover, the CSS images extracted from a database are processed and described by median vectors that constitutes the training data set for a SOM neural network. This way of description improves the accuracy of image retrieval in comparison with the previous work [1] that used the first principal component of the PCA technique. Experiments with a benchmark database are carried out to demonstrate the usefulness of the proposed methodology. Particle Detection on Electron Microscopy Micrographs Using Multi-Classifier Systems Lucas M. Oliveira, Raul B. Paradeda, Bruno M. Carvalho, Anne M. P. Canuto, and Marcilio C. P. de Souto Abstract: The determination of the three-dimensional (3D) structure of biological macromolecules at different configurations can be very important for understanding biological processes at the molecular level. The detection of individual particles from electron microscopy (EM) micrographs turns into a major la\-bor-intensive bottleneck, when the number of particles needed starts to exceed a few tens of thousand molecular images. Multi-classifier systems have been widely investigated as tools for performing complex classifying tasks. In this work, we investigate the adequacy of using multi-classifier systems to detect particles on electron microscopy micrographs. In order to do so, we compare the performance of five algorithms for generating individual classifiers and three other ones for multi-classifier algorithms. Such results are also compared with others found in the literature. In terms of results, the multi-classifier systems generated show larger accuracy (correct classification) and lower false positive and negative rates. Evolutionary Optimization of Wavelet Feature Sets for Real-Time Pedestrian Classification
Jan Salmen, Thorsten Suttorp, Johann Edelbrunner, and Christian Igel Abstract: Computer vision for object detection often relies on complex classifiers and large feature sets to achieve high detection rates. But when real-time constraints have to be met, for example in driver assistance systems, fast classifiers are required. We consider the design of a computationally efficient system for pedestrian detection. We propose an evolutionary algorithm for the optimization of a small set of wavelet features, which can be computed very efficiently. These features serve as input to a linear classifier. The classification performance of the optimized system is only slightly worse compared to recently published results obtained with support vector machines on large feature sets, while the computational time is lower by orders of magnitude. Session HMLII: Hybrid Metaheuristics and Learning, Tuesday, 18. Sept., 9.30–10.30. a.m., Z03.08 A New PSO Algorithm Incorporating Reproduction Operator for Solving Global Optimization Problems Millie Pant, Thangaraj Radha, and Ajith Abraham Abstract: in this paper we have presented a new variant of Basic Particle Swarm Optimization (BPSO) algorithm named QI-PSO for solving global optimization problems. The QI-PSO algorithm makes use of a multiparent, quadratic crossover/reproduction operator defined by us in the BPSO algorithm. We have compared it with Basic Particle Swarm Optimization and the numerical results show that QI-PSO outperforms the BPSO algorithm in all the sixteen cases taken in this study. Hybridized Swarm Metaheuristics for Evolutionary Random Forest Generation Miroslav Bursa, Lenka Lhotska, and Martin Macas Abstract: In many industry and research areas, data mining is a crucial process. This paper presents an evolving structure of classifiers (random forest) where the trees are generated by hybrid method combining Ant Colony metaheuristics and Evolutionary computing technique. The method benefits from the stochastic process and population approach, which allows the algorithm to evolve more efficiently than the methods alone. As the method is similar to random forest, it can be also used for feature selection. The paper also consults the parameter estimation for the method. Tests on real data (UCI and real biomedical data) have been performed and evaluated. The average accuracy of the method over MIT-BIH database with normalized data and equalized classes is sensitivity 93.22 % and specificity 87.13 %. Visualization of Pareto-Sets in Evolutionary Multi-Objective Optimization Mario Koeppen and Kaori Yoshida Abstract: In this paper, a method for the visualization of the population of an evolutionary multi-objective optimization (EMO) algorrithm is presented. The main characteristic of this approach is the preservation of Pareto-dominance relations among the individuals as good as possible. It will be shown that in general, a Pareto-dominance preserving mapping from higher- to lower-dimensional space does not exist, so the demand to have as few wrong dominance relations after the mapping as possible gives an objective in addition to other mapping objectives like preserving nearest neighbor relations. Thus, the mapping itself poses a multi-objective optimization problem by itself, which is also handled by an EMO algorithm (NSGA-II in this case). The resulting mappings are shown for the run of a modified NSGA-II on the 15 objective DTLZ2 problem as an example. From such plots, some insights into evolution dynamics can be obtained.
Special Session SIP: Aspects of Image Processing: Theory, Applications & Computations, Tuesday, 18. Sept., 11.00–12.30. a.m., Z02.02 Global Modes in Kernel Density Estimation: RAST Clustering Oliver Wirjadi and Thomas Breuel Abstract: The mean shift algorithm is a widely used method for finding local maxima in feature spaces. Mean shift algorithms have been shown in the literature to be equivalent to a gradient ascent optimization of a kernel density estimate. This paper describes a novel, globally optimal optimization method and compares the suboptimal mean shift solutions with the globally optimal solutions derived by the new algorithm. Experimental results on both simulated and real data show that new algorithm yields solutions that are often significantly better than the suboptimal solutions identified by the mean shift algorithm, and that it scales better to large sample sizes and is more robust to noise levels. Fiber Orientation Estimation from 3D Image Data: Practical Algorithms, Visualization, and Interpretation Katharina Robb, Oliver Wirjadi, and Katja Schladitz Abstract: Fibrous materials such as fiber-reinforced composites are finding increasing application in the automotive, aerospace, and other industries. Fiber arrangements and defects at microscopic scales have direct impact on their stability. Two-dimensional images obtained by either non-destructive or destructive imaging cannot reveal the full fiber orientation information. Therefore, three-dimensional images obtained by micro computed tomography ($\mu$CT) are used. Since segmentation of these image datasets is often difficult due to low contrast, we propose a linear filtering scheme to extract local fiber orientations. Efficient implementations of these filters have been proposed, resulting in practical algorithms with acceptable runtimes in the scale of minutes to at most a few hours for common tomographic image sizes. We show how to condense the local orientation information into visual representations. In contrast to existing 3D orientation estimation methods, our method results in densely sampled orientation maps. The proposed method is applied to images of two different fiber materials and compared to orientation estimates based on measures obtained from integral geometry. We show conformance of the proposed orientation estimation methods with these known methods. A Hybrid Texture Analysis System based on Non-Linear & Oriented Kernels, Particle Swarm Optimization and kNN vs. Support Vector Machines Stefanie Peters and Andreas König Abstract: This paper expands our previous activities on automatically texture analysis applying optimized non-linear and oriented kernels. The operator parameterization is achieved using particle swarm optimization (PSO). The sensitivity of the voting kNN classifier used in the optimization process and for texture classification is explored in respect of the number of used neighbors. Additionally, support vector machines (SVM) with the reputation to procure better results are applied. Contrary to a recommended grid search for the parameter selection, the adaptation of the free SVM parameters is included into the global optimization process with PSO. Our work was tested with benchmark and application data from leather inspection.
Session HMI: Hybrid Models I, Tuesday, 18. Sept., 11.00–12.20. a.m., Z03.08 Application of a Hybrid Classifier to the Recognition of Petrochemical Odors Eleonora Oliveira, Paulemir Campos, and Teresa Ludermir Abstract: Nowadays there are several data mining algorithms applied to the resolution of many different problems, such as the classification of patterns. However, when these algorithms are used separately to classify they usually present an inferior performance compared to the performance obtained by combined models. The Bagging and Boosting techniques combine models of the same kind in a competitive form, in other words, the output is generally provided by the winning classifier. Alternatively, Stacking usually combines different algorithms, constituting a hybrid model. Nevertheless, stacking has a high cost, due to the search for the best models that will be combined to solve a certain problem. Thus, we present a Hybrid Classifier (HC) to be applied to the recognition of gases derived from petrol at a lower cost and in a cooperative way. An Immune Genetic Algorithm for Software Test Data Generation Abdel Hamid Bouchachia Abstract: This paper aims at incorporating immune operators in genetic algorithms as an advanced method for solving the problem of test data generation. The new proposed hybrid algorithm is called Immune Genetic Algorithm (IGA). A full description of this algorithm is presented before investigating its application in the context of software test data generation using some benchmark programs. Moreover, the algorithm is compared with other evolutionary algorithms. Fuzzy Neural Network Model Using a Fuzzy Learning Vector Quantization Yong Soo Kim and Sung-ihl Kim Abstract: In this paper, we propose a fuzzy LVQ(Learning Vector Quantization) which is based on the fuzzification of LVQ. The proposed fuzzy LVQ uses the different learning rate depending on whether classification is correct or not. When the classification is correct, it uses the combination of a function of the distance between the input vector and the prototypes of classes and a function of the number of iteration as the fuzzy learning rate. On the other hand, when the classification is not correct, it uses the combination of the fuzzy membership value and a function of the number of iteration as the fuzzy learning rate. The proposed fuzzy LVQ is integrated into the supervised IAFC(Integrated Adaptive Fuzzy Clustering) neural network 5. We used iris data set to compare with performance of the supervised IAFC neural network 5 with those of LVQ algorithm and backpropagation neural network. The supervised IAFC neural network 5 yielded fewer misclassifications than LVQ algorithm and backpropagation neural network. Evolutionary Approaches to Solve an Integrated Lot Scheduling Problem in the Soft Drink Industry Claudio Toledo, Paulo França, Reinaldo Morabito, and Alf Kimms Abstract: This paper proposes two evolutionary approaches as procedures to solve the Synchronized and Integrated Two-Level Lot-Sizing and Scheduling Problem (SITLSP). This problem can be found in some industrial settings, mainly soft drink companies, where the production process involves two interdependent levels with decisions concerning raw material storage and soft drink bottling. The first approach to solve the SITLSP is a Multi-Population Genetic Algorithm (GA) with a hierarchical ternary tree structure for populations. The second approach is a Me-metic Algorithm (MA) that extends the GA approach through the inclusion of a local search procedure. The computational study reported reveals that those methods are an effective alternative to solve real-world instances of the SITLSP.
Session HMII: Hybrid Models II, Tuesday, 18. Sept., 2.30–3.30. p.m., Z02.02 Bayesian Networks Modeling for Asian Suprema Soybean Rust Incidence Study in Different Conditions of Temperatures and Leaf Wetness Ricardo Martins de Abreu Silva, Felipe L. Valentim, and Marcelo C. Alves Abstract: The Asian soybean rust (Phakopsora pachyrhizi H. Sydow & P. Sydow), which has been reported in areas of tropical and subtropical climates around the world, causes significant soybean (Glycine max L. Merr.) yield reduction. The disease progress is influenced by biotic factors as interaction pathogen/host and abiotic factors of the environment. This work presents three models based on bayesian network to study of Asian Suprema soybean rust incidence in different temperature and leaf wetness conditions. The models present estimates equivalents to non-linear regression model (Reis et al., 2004) fuzzy model (Alves et al., 2006) and neuro-fuzzy model (Silva et al., 2006), when compared on the results from experimental design realized by (Alves et al., 2004). Hybrid Approach to Solve a Crew Scheduling Problem: an Exact Column Generation Algorithm Improved by Metaheuristics André Santos and Geraldo Mateus Abstract: This paper shows a successful hybrid approach to improve a column generation algorithm. The objective is to construct daily duties to bus drivers, in order to cover a set of trips. Due to a large number of variables, the problem is decomposed in a master and a subproblem. The subproblem iteratively generates duties to the master problem, so the main task is to solve the subproblem. An exact ILP model may do this, but it is generally time consuming. We propose a heuristic based in the linear relaxation of this model to quickly generate many duties, and the ILP is called just when the heuristic fails, so it is possible to obtain and prove optimality. We also use two metaheuristics to solve the subproblem: GRASP and genetic algorithm. All three heuristics improves the column generation algorithm and a hybrid approach using two of them turns out to be even faster. Model and Algorithms for the Multicommodity Traveling Salesman Problem João Sarubbi, Geraldo Mateus, Luna Henrique, and Miranda Jr. Gilberto Abstract: Suppose that a traveling salesman has to deliver a quantity dk of a specific commodity to each node k of a network. The traveling salesman pays the standard fixed cost to pass in an arc$(i,j)$, but he also faces a variable cost for each kind of commodity that needs to be carried across that arc. Each variable cost is proportional to the quantity of the correspondent commodity, in such a way that differences in both the unitary arc costs and the quantities to be delivered imply in differences on the operational costs to serve the customers nodes. We are presenting another formulation for this TSP related problem with a Lagrangean Algorithm that finds good lower bounds even for instances with solver CPLEX is not able to run. Session HSPII: Hybrid Signal Processing II, Tuesday, 18. Sept., 2.30–3.30. p.m., Z03.08 Content Based Image Retrieval using a Descriptors Hierarchy Raquel Esperanza Patiño Escarcina and Jose Alfredo Ferreira Costa
Abstract: Content based image retrieval tries to find a set of images similar to a given example. Low level descriptors can be used to represent and index images, but the gap between this descriptors and abstract concepts is the main problem in CBIR and it is necessary to propose new algorithms to bridge this problem. In the other hand, objects can be found in a collection by looking general features and discarding those objects that do not fit into these features to reduce the search space. Next, other more specific feature can help to find these objects in the reduced search space. This approach is inspired in the former idea and propose the arrangement of low level descriptors into a hierarchy. This arrangement has to be done considering detail of information each descriptor gives. Finally, descriptors on each level of the hierarchy are used to index images in the search space and a filter to reduce it has to be execute. This process is repeated until the low level of the hierarchy is reached. Experiments demonstrate the effectiveness of the proposed approach compared with the traditional ones and reveal it as a good option to implement CBIR systems. Pareto-dominated Hypervolume Measure: An Alternative Approach to Color Morphology Mario Koeppen and Katrin Franke Abstract: In this paper, an alternative approach to the non-linear filtering of Mathematical Morphology for color and multispectral images based on the Pareto-dominated Hypervolume measure is presented. Pareto-set theory is studied in this context, and among others, successfully implemented in the morphological filtering of color images. The demand to assess the quality of a multi-objective optimization algorithms, in particular, has put forth several kind of measures. One of these measures, the Hypervolume, bases on the Lebesque measure of all points dominated by a set of points and maps a set of Pareto-optimal points to a scalar. By considering the Hypervolume in the color-image domain, it can be shown that this Hypervolume corresponds to another kind of Color Morphology, were each pixel in the filtered image represents the Hypervolume of its set of neighbours in the original color image. In the following, some properties of the Hypervolume, as used as an image processing filter will be derived, and some potential applications of this approach to Color Morphology will be shown. A Novel Fully Evolved Kernel Method for Feature Computation from Multisensor Signal Using Evolutionary Algorithms Kuncup Iswandy and Andreas König Abstract: The design of intelligent sensor systems requires sophisticated methods from conventional signal processing and computational intelligence. Currently, a significant part of the overall system architecture still has to be manually elaborated in a tedious and time consuming process by an experienced designer for each new application or modification. Clearly, an automatic method for auto-configuration of sensor systems would be salient. In this paper, we contribute to the optimization of the feature computation step in the overall system design, investigating Gaussian kernel methods. Our goal is to improve the kernel method of feature computation with consideration on including the adjustable magnitude parameter for Gaussian kernels or fully evolved Gaussian kernels, which are inspired by feature weighting concepts and are similar to RBF like neural network with correlation based kernel layer and linear combiner output layer. We will compare this improved method with previous kernel methods using multiobjective evolutionary optimization, i.e., genetic algorithms. In addition to the straightforward feature space from the optimized kernel layer, we complement the kernel layer by linear combiner layer, with weights computed by traditional LDA (linear discriminant analysis) in the loop of the optimization. In our experiments we applied gas sensor benchmark data and the results showed that our novel method can achieve competitive or even better recognition accuracies and effectively reduce the computational complexity as well. Postersession, Tuesday, 18. Sept., 4.00–5.30. p.m., Foyer
Mining Medical Databases using Incremental Enhanced Association Rules Algorithm Laila Elfangary and Walid Atteya Abstract: Maintenance of association rules is an important problem. When new transactions are added to the set of old transaction database, how can we update the association rules already discovered in the set of old transactions efficiently? This paper presents an enhanced algorithm for mining incremental updates in large databases. Our paper shows that the algorithm performs significantly faster than the approach of mining the whole updated database from scratch. We first present a previously implemented algorithm namely the Proposed Enhanced Algorithm for mining association rules in large databases. Next, we propose our algorithm for mining incremental updates in the database and efficiently updating the discovered association rules. Finally, we present some scale up experiments that show how the Proposed Incremental Algorithm outperforms the Proposed Enhanced Algorithm. Graphical charts are used to visualize and to help in interpreting the statistical significance of the results. The goal is to present how methods and tools for intelligent data analysis are helpful in narrowing the gap between data gathering and data analysis. The PCR Primer Design as a Metaheuristic Search Process Luciana Montera and Maria do Carmo Nicoletti Abstract: The Polymerase Chain Reaction process is a well-known technique for the in vitro amplification of a DNA sequence. The success of a PCR depends on several parameters particularly the primer sequences used. Since the design of a suitable pair of primer involves a reasonable number of variables, which can have a range of different values, computer programs are commonly used to assist this task. This paper approaches the design of a pair of primer sequences as a search process throughout the space defined by all possible primer sequence pairs. In order to do this, a simulated annealing based search strategy was implemented and an experiment is discussed. A Study of Classification Algorithms for Data Mining Based on Hybrid Intelligent Systems Wang Gang, Huang LiHua, and Zhang ChengHong Abstract: Facing the huge amounts of data, the familiar classification algorithms show the shortages on time efficiency, robustness and accuracy. So this article puts the Hybrid Intelligent Systems into the research of classification algorithm. Based on the cognitive psychology and aggregative model theory, the article proposes a new Hybrid Intelligent Systems: R-FC-DENN, according to Rough Set, Clustering theory, Fuzzy Logic, Genetic Algorithm and Artificial Neural Network. Firstly, R-FC-DENN uses the Rough Set to reduce the data. And then it clusters the data by the Clustering theory. After that, it uses different and improved ANN to train. Subsequently, the data which are trained are integrated by fuzzy power. Lastly, the integrated data are trained by another improved ANN and the whole process of training is completed. In the end, experiments are carried out based on the data of UCI database and it is observed that the system is valid. Genetic Programming meets Model-Driven Development Thomas Weise, Michael Zapf, Mohammad Ullah Khan, and Kurt Geihs Abstract: Genetic programming is known to provide good solutions for many problems like the evolution of network protocols and distributed algorithms. In such cases it is most likely a hardwired module of a design framework that assists the engineer to optimize specific aspects of the system to be developed. It provides its results in a fixed format through an internal interface. In this paper we show how the utility of genetic programming can be increased remarkably by isolating it as a component and integrating it into the model-driven software development process. Our genetic programming framework produces XMI-encoded UML models that can easily be loaded into widely available modeling tools which in turn offer code generation as well as
additional analysis and test capabilities. We use the evolution of a distributed election algorithm as an example to illustrate how genetic programming can be combined with model-driven development. This example clearly illustrates the advantages of our approach - the generation of source code in different programming languages. Particle Swarm Optimization of Neural Network Architectures and Weights Marcio Carvalho and Teresa Ludermir Abstract: The optimization of architecture and weights of feed forward neural networks is a complex task of great importance in problems of supervised learning. In this work we analyze the use of the Particle Swarm Optimization algorithm for the optimization of neural network architectures and weights aiming better generalization performances through the creation of a compromise between low architectural complexity and low training errors. For evaluating these algorithms we apply them to benchmark classification problems of the medical field. The results showed that a PSO-PSO based approach represents a valid alternative to optimize weights and architectures of MLP neural networks. Generating Fuzzy Rules from Examples Using the Particle Swarm Optimization Ahmed Esmin Abstract: The use of Fuzzy Logic to solve control problems have been increasing considerably in the past years. The problem of generating desirable fuzzy rules is very important in the development of fuzzy systems. This paper presents a generation method of fuzzy rule by learning from examples using the Particle Swarm Optimization method (PSO). The proposed algorithm can obtain a set of fuzzy rules which cover the examples set in iterative process. Dataflow Orchestration of Image Processing Algorithms Using High Level Petri Nets Björn Wagner, Andreas Dinges, and Paul Müller Abstract: Image processing algorithms for industrial systems tend to be very complex and specialised to the particular case. We present a new way of modeling image-processing algorithms for distributed systems using high level petri-nets. We present the modeling of image-processing as dataflow networks of basic building blocks. Petri nets seem to be well suited for modeling such dataflow networks. We analyse the the usability and limitations of classical petri nets for this task and go on into the required extensions. Furthermore we present our implementation of a distributed service-oriented industrial image-processing system. Learning Context-Free Grammars from Partially Structured Examples: Juxtaposition of GCS with TBL Olgierd Unold and Lukasz Cielecki Abstract: This paper juxtaposes performance of the grammar-based classifier system (GCS) with tabular representation algorithm (TBL) on the task of inducing context-free grammars from partially structured examples. In both cases structured examples rapidly improve the efficiency of learning algorithms, although there are substantial differences in working among them. GCS requires more attention while setting initial system parameters and structured examples provide a bit more information than in TBL but on the other hand it is more efficient when tested against the same examples sets. Software Effort Estimation using Machine Learning Techniques with Robust Confidence Intervals
Petronio Braga, Adriano Oliveira, and Silvio Meira Abstract: The precision and reliability of the estimation of the effort of software projects is very important for the competitiveness of software companies. Good estimates play a very important role in the management of software projects. Most methods proposed for effort estimation, including methods based on machine learning, provide only an estimate of the effort for a novel project. In this paper we introduce a method based on machine learning which gives the estimation of the effort together with a confidence interval for it. In our method, we propose to employ robust confidence intervals, which do not depend on the form of probability distribution of the errors in the training set. We report on a number of experiments using two datasets aimed to compare machine learning techniques for software effort estimation and to show that robust confidence intervals can be successfully built. Fuzzy and Neuro-Fuzzy Modeling for Total Volume study of Eucalyptus sp Ricardo Martins de Abreu Silva, Adriano R. de Mendonça, Fillipe G. Brandão, Danilo M. Pires, and Gleimar B. Baleeiro Abstract: Eucalyptus is the most valuable cultivated forest genus in Brazil nowadays. Modeling eucalypts volume has been important for foresters in recent years due to strong site and genetic variations, management regimes and multiple products generated from these plantations. At this work, fuzzy and neuro-fuzzy models were developed to represent the volume pattern of eucalypts clonal stands from Brazilian coast region. Likewise in other scientific field, these types of modeling methodologies showed to be precise and accurate. Dynamic Overlay Networks for Image Processing Grids Andreas Dinges, Bjoern Wagner, and Paul Mueller Abstract: During the development and parametrization of 2D image-processing algorithms for surface inspection uses, you need to test a huge amount of image-data for each modification of the algorithms or parameters. For algorithm runtimes up to several seconds, this will take a long time. To speed up this process it is recommended to distribute the computation in a parallel computation environment. Compute Grids, which use the unused resources of existing hardware are the most cost efficient way to solve this problem. The most existing Grid-Concepts are based on flat connection structures with a scheduler on the top; for high job-rates the scheduler will become the bottleneck of the whole system. Concepts to solve this problem organize the nodes in tree-structures to discharge the central scheduler. In heterogeneous Desktop-Grids where the different nodes are widely distributed the normaly used random arrangement of the nodes in the tree-structure can be counterproductive, because the bandwidths and latencies in a Grid are varying. In this paper there is shown a solution to arrange the nodes of the grid optimized by bandwidth and latency, using modified spanning-tree algorithms, so that the average response time is reduced and from this following the job-throughput of the Compute-Grid is increased. Exploration of Pareto Frontier Using a Fuzzy Controlled Hybrid Line Search Crina Grosan and Ajith Abraham Abstract: This paper proposes a new approach for multicriteria optimization which aggregates the objective functions and uses a line search method in order to locate an approximate efficient point. Once the first Pareto solution is obtained, a simplified version of the former one is used in the context of Pareto dominance to obtain a set of efficient points, which will assure a thorough distribution of solutions on the Pareto frontier. In the current form, the proposed technique is well suitable for problems having multiple objectives (it is not limited to bi-objective problems) and require the functions to be continuous twice differentiable. In order to assess the effectiveness of this approach, some experiments were performed and compared with two recent well known population-based meta-heuristics ParEGO [11] and NSGA II [2]. When compared to ParEGO and NSGA II, the proposed approach not only assures a better convergence to the Pareto frontier but also illustrates a good distribution of solutions. From a computational point of view, both stages of the line search converge within a short time (average about 150 milliseconds for the first stage and about 20 milliseconds for the second stage). Apart from this, the proposed technique is very simple, easy to implement to solve multiobjective problems.
Comparative Study of Clustering Techniques for the Organization of Software Repositories Ronaldo Veras, Adriano Oliveira, Bruno Melo, and Silvio Meira Abstract: Software reuse is essential for improving the productivity and quality of software projects. One of the key issues to promote the adoption of software reuse in companies is the development of effective repositories of software components. It is also very important to have good methods for searching and retrieval of the components. Clustering techniques can help by providing a visualization of the repository of software components as well as in helping to refine the searches by grouping together similar components. In this paper we quantitatively compare two clustering techniques, namely, self-organizing maps (SOM) and growing hierarchical SOM (GHSOM) for clustering a repository of classes from a Java API for building mobile systems. The performance measure was the quantization error. The simulations have shown that GHSOM outperforms SOM in these tasks. GHSOM is more suitable for this task because it is a constructive technique, which is an advantage in tackling the growth of the repository of software components. Special Session AMI: Ambient Intelligence [at Home and Recreation], Wednesday, 19. Sept., 10.00–12.00. a.m., Z02.02 Sensor-based Training Optimization of a Cyclist Group Ankang Le, Thomas Jaitner, and Lothar Litz Abstract: Determining the optimal exercise intensity is a crucial factor to increase performance in professional cycling. Novel sensor technologies allow to optimize the training not only for an individual cyclist but also for an entire training group. A sensor-based Assisted Bicycle Trainer (ABT) system with a control algorithm has been developed at the University of Kaiserslautern to optimize the group training in cycling. The focus of this paper is on the development of the control algorithm, a Model Predictive Controller (MPC) for the optimization of the group training. The controller predicts the heart rate of the cyclists based on individualized heart rate models and regulates the group training by advising cyclists to change the order of the group, to adjust the group speed, or to split the group in such a way that each cyclist can meet his training plan as exactly as possible. Indoor Localisation of Humans, Objects, and Mobile Robots with RFID Infrastructure Jan Koch, Jens Wettach, Eduard Bloch, and Karsten Berns Abstract: The need for robust indoor localisation for all types of entities has been under continuous research by the ubiquitous community. Intelligent environments have to be supported with contextual information in order to facilitate intelligent behaviour. These contextual information include the location of humans and objects within the particular environment. Intelligent environments can be living areas with home automation, smart industrial plants, sensor-equipped office areas and indoor-emergency applications. So far technical solutions are either quite expensive or lack of precision for robust usage as components in intelligent service federations. We present rather low-cost localisation systems with great scalability based on active and passive RFID technology to locate humans, mobile service robots and objects of the daily use. The trade-off between technical effort and costs on the one hand and sufficient data accuracy for the application on the other hand is discussed. A motivation of our scenario, the technical concept and solution as well as the implementation and the integration that so far have been performed will be presented. Current prototypes of the proposed system are already being tested in a project aiming on development of smart assisted living environments. MacZ - A Quality-of-Service MAC Layer for Ad-hoc Networks Philipp Becker, Reinhard Gotzhein, and Thomas Kuhn
Abstract: We present a new MAC layer called MacZ, specifically devised for flexible QoS support in mobile ad-hoc networks. MacZ provides multi-hop relative time synchronization, network-wide medium slotting, contention-free transmission with global reservations, contention-based transmission with priorities, and is robust against topology changes. We describe the main functionalities of MacZ, show results of performance simulations that provide evidence for the effectiveness of the QoS mechanisms, and discuss related MAC layers. A Hybrid Approach to Intelligent Living Assistance Markus Nick and Martin Becker Abstract: IT-based living assistance systems focusing on the support of people with special needs in their daily routine have to continuously monitor and assist them in an appropriate way. To this end, we have developed a Monitoring and Assistance component including a hybrid reasoner that is able to adapt planned and running treatments according to the current situation and context. In this paper, we the explain the underlying approaches followed in the reasoner, describe the reasoning technologies used for this task and its sub-tasks, and present some first evaluation results. Special Session ENNH: Embedded Neural Network Hardware, Wednesday, 19. Sept., 10.00–12.00. a.m., Z03.08 Neighborhood Rank Order Coding for Robust Texture Analysis and Feature Extraction Christian Mayr and Rene Schüffny Abstract: Research into the visual cortex and general neural information processing has led to various attempts to integrate pulse computation schemes in image analysis systems. Of interest is especially the robustness of representing an analogue signal in the phase or duration of a pulsed, quasi-digital signal, as well as the possibility of direct digital interaction, i.e. computation, among these signals. Such a computation can also achieve information compaction for subsequent processing stages. By using a pulse order encoding scheme motivated by dendritic pulse interaction, we will show that a powerful low-level feature and texture extraction operator, called Pulsed Local Orientation Coding (PLOC), can be implemented. Feature extraction results are being presented, and a possible VLSI implementation is detailed. A Neuromorphic aVLSI Network Chip With Configurable Plastic Synapses Patrick Camilleri, Massimiliano Giulioni, Vittorio Dante, Davide Badoni, and Giacomo Indiveri Abstract: We describe and demonstrate a neuromorphic, analog VLSI chip (termed F-LANN) hosting 128 integrate-and-fire (IF) neurons with spike-frequency adaptation, and 16,384 plastic bistable synapses implementing a self-regulated form of Hebbian, spike-driven, stochastic plasticity. The chip is designed to offer a high degree of reconfigurability: each synapse may be individually configured at any time to be either excitatory or inhibitory and to receive either recurrent input from an on-chip neuron or AER-based input from an off-chip neuron. The initial state of each synapse can be set as potentiated or depressed, and the state of each synapse can be read and stored on a computer A Digital Framework for Pulse Coded Neural Network Hardware with Bit-Serial Operation Tim Kaulmann, Deniz Dikmen, and Ulrich Rueckert Abstract: This publication presents a digital framework for builing up pulse coded neural networks with leaky integrate-and-fire neurons and static synapses as well as dynamic synapses. The system, including a novel
communication infrastructure, is mainly focused on ASIC systhesis but also shows a small footprint on Virtex2(Pro) FPGAs. Its bit-serial operation has been verified in simulations. Hybrid Intelligent and Adaptive Sensor Systems with Improved Noise Invulnerability by Dynamically Reconfigurable Matched Sensor Electronics Senthil Kumar Lakshmanan and Andreas König Abstract: Hybrid intelligent sensor systems and networks are composed of modules of tightly co-operating software and hardware components. Bio-inspired information processing is embodied in algorithms as well as dedicated electronics for intelligent processing and system adaptation. This paper focuses on the challenges imposed on the small yet irreplaceable analog and mixed signal components in such a sensor system, which are prone to deviation and degradations. Novel architectures combine issues of rapid-prototyping, trimming, fault-tolerance, and self-repair. However, the common reconfiguration approaches cannot deal efficiently with real-world noise problems. This paper adapts effective solution strategies to advanced sensor electronics for hybrid intelligent and adaptive sensor systems in a 0.35µm CMOS techno-logy and reports on the design of a novel generic chip. Session HIPII: Hybrid Information Processing II, Wednesday, 19. Sept., 2.00 – 3.20. p.m., Z02.02 How to Obtain Fair Managerial Decisions in Sugar Cane Harvest Using NSGA-II Diogo Pacheco, Tarcísio Lucas, and Fernando Lima Neto Abstract: The world’s demand for sugar and particularly for renewable fuels such as ethanol requires an increase in production in sugar mills. The use of artificial neural networks (ANN) posed as a predictive core associated with the algorithm NSGA-II aims at helping decision makers to optimize the multi-objective harvest problem. This paper presents two approaches and the good results achieved as compared with other classical techniques. Markov-Blanket Based Strategy for Translating a Bayesian Classifier into a Reduced Set of Classification Rules Estevam Hruschka Jr., Maria Nicoletti, Vilma Oliveira, and Glaucia Bressan Abstract: Bayesian network (BN) is a formalism for representing and reasoning about uncertain domains. In BN the knowledge is represented by a combination of a graph-based structure and probability theory. A particular type of BN known as Bayesian Classifier (BC) aims at classifying a given instance into a discrete class. BCs have been extensively used for modeling knowledge in many different applications and have been the focus of many works related to data mining. Depending on the size of a BC the understandability of the knowledge it represents is not an easy task. This paper proposes an approach to help the process of understanding the knowledge represented by a BC, by translating it into a more convenient and easily understandable form of representation, that of classification rules. The proposed method named BayesRule (BR) uses the concept of Markov Blanket (MB) to obtain a reduced set of rules in respect to both, the number of rules and the number of antecedents in rules. Experiments using the ALARM network showed that the reduced set of rules extracted from the BC can be smaller than the set of rules representing a decision tree generated by C4.5, and still maintains a high accuracy rate. Learning to Reach Optimal Equilibrium by Influence of Other Agents Opinion Dennis Barrios Aranibar and Luiz Marcos Garcia Gonçalvez
Abstract: In this work we propose a new paradigm for learning coordination in multi-agent systems. This approach is based on social interaction of people, specially in the fact that people communicate to each other what they think about their actions and this opinion has some influence in the behavior of each other. We propose a model in which multi-agents learn to coordinate their actions giving opinions about the actions of other agents and also being influenced with opinions of other agents about their actions. We use the proposed paradigm to develop a modified version of the Q-learning algorithm. The new algorithm is tested and compared with independent learning (IL) and joint action learning (JAL) in a grid problem with two agents learning to coordinate. Our approach shows to have more probability to converge to an optimal equilibrium than IL and JAL Q-learning algorithms, specially when exploration increases. Also, a nice property of our algorithm is that it does not need to make an entire model of all joint actions like JAL algorithms. Variable Ordering in the Conditional Independence Bayesian Classifier Induction Process: An Evolutionary Approach Estevam Hruschka Jr., Edimilson Santos, and Sebastian Galvao Abstract: This work proposes, implements and discusses a hybrid Bayes/GA collaboration (VOGAC-MarkovPC) designed to induce Conditional Independence Bayesian Classifiers from data. The main contribution is the use of MarkovPC algorithm in order to reduce the computational complexity of a Genetic Algorithm (GA) designed to explore the Variable Orderings (VOs) in order to optimize the induced classifiers. Experiments performed in a number of datasets revealed that VOGAC-MarkovPC required less than 25% of the time demanded by VOGAC-PC on average. In addition, when concerning the classification accuracy, VOGAC-MakovPC performed as well as VOGAC-PC did. Session DFN: Interdisciplinary/Hybrid Approaches to Dependable and Flexible Networking, Wednesday, 19. Sept., 2.00 – 3.30. p.m., Z03.08 Performance Trade-off Exploration by Query-Trail-Mediated Topology Reconstruction in Unstructured P2P Networks Kei Ohnishi, Satoshi Nagamatsu, and Yuji Oie Abstract: This paper presents a topology reconstruction method to explore better trade-off points between search and access load balancing performance in unstructured Peer-to-Peer (P2P) networks for file sharing. The proposed topology reconstruction method changes a network topology in dynamic, autonomic, and decentralized manner. The topology reconstruction is based on local threshold-based rules, and these rules utilize query trails that mean information on the previous successful search paths. A power-law network is used as the initial network in simulations. The simulation results show that depending on setting of the threshold values, the proposed method can explore better trade-off points between search and storage access load balancing performance compared to the case of not doing topology reconstruction. A Control Method of a P2P Network with Small Degree and Diameter Yusuke Sasaki and Hiroyoshi Miwa Abstract: In this paper, we propose a control method of a P2P network with its maximum degree 4 for routing and its diameter log2 n where n is the number of nodes. In previous methods, the maximum degree or the diameter is only probabilistically bounded. As the degree of a vertex is the number of the neighbor nodes, a node with a large degree suffers high load to transfer many queries. Even if the average degree is bounded, the loads of some nodes with large degrees are exponentially high. As the diameter of a network is the worst distance, a large diameter causes bad response time. Even if the average diameter is bounded, the performance between two nodes apart from each other is always bad. Therefore, it is important that the degrees and the diameter are always
small. We show that the proposed method has this good property. Implementation and Experimental Evaluation of On-Line Simulation Server for OSPF-TE Hitomi Tamura, Tsuyoshi Okubo, Yousuke Inoue, Kenji Kawahara, and Yuji Oie Abstract: As the amount of traffic transfered on the Internet are increasing, dynamic Traffic Engineering (TE) becomes important to avoid link congestion. In Open Shortest Path Fast(OSPF)-based networks, link costs are statically set according to its long-term utilization for reducing traffic on some congested nodes, hence temporary performance degradation may occur due to short-term traffic fluctuation. For dynamic traffic engineering on OSPF-based networks, measurement of the utilization of links / nodes, inferring the set of link costs for improving transmission behavior and setting the cost set to routers are necessary and so-called On-Line Simulation (OLS) system can operate these functions autonomously and periodically. In this paper, we construct the server prototype in OLS system, and evaluate its scalability and control performance in our testbed network. Experimental results show that the server succeeds in providing low-cost network management and real-time control even if there is the large amount of traffic on the network. Furthermore, the total throughput over the network was greatly improved by the OLS.