Intelligent and Evolutionary Systems

217 Pages·2009·7.85 MB·English

by  Yukiko Orito

MitsuoGen,DavidGreen,OsamuKatai,BobMcKay,AkiraNamatame,
RuhulA.Sarker,andByoung-TakZhang(Eds.)
IntelligentandEvolutionarySystems
StudiesinComputationalIntelligence,Volume187
Editor-in-Chief
Prof.JanuszKacprzyk
SystemsResearchInstitute
PolishAcademyofSciences
ul.Newelska6
01-447Warsaw
Poland
E-mail:[email protected]
Furthervolumesofthisseriescanbefoundonourhomepage: Vol.176.BeniaminoMurgante,GiuseppeBorrusoand
springer.com AlessandraLapucci(Eds.)
GeocomputationandUrbanPlanning,2009
ISBN978-3-540-89929-7
Vol.165.DjamelA.Zighed,ShusakuTsumoto,
ZbigniewW.RasandHakimHacid(Eds.) Vol.177.DikaiLiu,LingfengWangandKayChenTan(Eds.)
MiningComplexData,2009 DesignandControlofIntelligentRoboticSystems,2009
ISBN978-3-540-88066-0 ISBN978-3-540-89932-7
Vol.166.ConstantinosKoutsojannisandSpirosSirmakessis Vol.178.SwagatamDas,AjithAbrahamandAmitKonar
(Eds.) MetaheuristicClustering,2009
ToolsandApplicationswithArtificialIntelligence,2009 ISBN978-3-540-92172-1
ISBN978-3-540-88068-4
Vol.179.MirceaGh.NegoitaandSorinHintea
Vol.167.NgocThanhNguyenandLakhmiC.Jain(Eds.) Bio-InspiredTechnologiesfortheHardwareofAdaptiveSystems,
IntelligentAgentsintheEvolutionofWebandApplications,2009 2009
ISBN978-3-540-88070-7 ISBN978-3-540-76994-1
Vol.168.AndreasTolkandLakhmiC.Jain(Eds.) Vol.180.WojciechMitkowskiandJanuszKacprzyk(Eds.)
ComplexSystemsinKnowledge-basedEnvironments:Theory, ModellingDynamicsinProcessesandSystems,2009
ModelsandApplications,2009 ISBN978-3-540-92202-5
ISBN978-3-540-88074-5
Vol.181.GeorgiosMiaoulisandDimitriPlemenos(Eds.)
Vol.169.NadiaNedjah,LuizadeMacedoMourelleand IntelligentSceneModellingInformationSystems,2009
JanuszKacprzyk(Eds.) ISBN978-3-540-92901-7
InnovativeApplicationsinDataMining,2009
ISBN978-3-540-88044-8 Vol.182.AndrzejBargielaandWitoldPedrycz(Eds.)
Human-CentricInformationProcessingThroughGranular
Vol.170.LakhmiC.JainandNgocThanhNguyen(Eds.) Modelling,2009
KnowledgeProcessingandDecisionMakinginAgent-Based ISBN978-3-540-92915-4
Systems,2009
Vol.183.MarcoA.C.PachecoandMarleyM.B.R.Vellasco(Eds.)
ISBN978-3-540-88048-6
IntelligentSystemsinOilFieldDevelopmentunderUncertainty,
Vol.171.Chi-KeongGoh,Yew-SoonOngandKayChenTan 2009
(Eds.) ISBN978-3-540-92999-4
Multi-ObjectiveMemeticAlgorithms,2009
Vol.184.LjupcoKocarev,ZbigniewGaliasandShiguoLian
ISBN978-3-540-88050-9
(Eds.)
Vol.172.I-HsienTingandHui-JuWu(Eds.) IntelligentComputingBasedonChaos,2009
WebMiningApplicationsinE-CommerceandE-Services,2009 ISBN978-3-540-95971-7
ISBN978-3-540-88080-6
Vol.185.AnthonyBrabazonandMichaelO’Neill(Eds.)
Vol.173.TobiasGrosche NaturalComputinginComputationalFinance,2009
ComputationalIntelligenceinIntegratedAirlineScheduling, ISBN978-3-540-95973-1
2009
Vol.186.Chi-KeongGohandKayChenTan
ISBN978-3-540-89886-3
EvolutionaryMulti-objectiveOptimizationinUncertain
Vol.174.AjithAbraham,RafaelFalco´nandRafaelBello(Eds.) Environments,2009
RoughSetTheory:ATrueLandmarkinDataAnalysis,2009 ISBN978-3-540-95975-5
ISBN978-3-540-89886-3
Vol.187.MitsuoGen,DavidGreen,OsamuKatai,BobMcKay,
Vol.175.GodfreyC.OnwuboluandDonaldDavendra(Eds.) AkiraNamatame,RuhulA.SarkerandByoung-TakZhang
DifferentialEvolution:AHandbookforGlobal (Eds.)
Permutation-BasedCombinatorialOptimization,2009 IntelligentandEvolutionarySystems,2009
ISBN978-3-540-92150-9 ISBN978-3-540-95977-9
Mitsuo Gen
David Green
Osamu Katai
Bob McKay
Akira Namatame
RuhulA.Sarker
Byoung-Tak Zhang
(Eds.)
Intelligent and Evolutionary
Systems
123
MitsuoGen BobMcKay
WasedaUniversity SchoolofComputerScienceandEngineering
GraduateSchoolofIPS SeoulNationalUniversity
2-8Hibikino Gwanangno599
Wakamatsu-ku,Kitakyushu808-0135 Seoul151-744,Korea
Japan E-mail:[email protected]
E-mail:[email protected]
AkiraNamatame
DavidGreen Dept.ofComputerScience
ClaytonSchoolofInformationTechnology NationalDefenseAcademyofJapan
MonashUniversity Yokosuka,239-8686,Japan
ClaytonVictoria3800,Australia E-mail:[email protected]
E-mail:[email protected]
RuhulSarker
OsamuKatai SchoolofIT&EE,UNSW@ADFA
NorthcottDve
Dept.ofSystemsScience
Campbell,ACT2600,Australia
GraduateSchoolofInformatics
E-mail:[email protected]
KyotoUniversity
Sakyo-ku,Kyoto606-8501,Japan Byoung-TakZhang
E-mail:[email protected]
SchoolofComputerScienceandEngineering
SeoulNationalUniversity
Gwanangno599
Seoul151-744,Korea
E-mail:[email protected]
ISBN978-3-540-95977-9 e-ISBN978-3-540-95978-6
DOI10.1007/978-3-540-95978-6
StudiesinComputationalIntelligence ISSN1860949X
LibraryofCongressControlNumber:2008944016
(cid:1)c 2009Springer-VerlagBerlinHeidelberg
This work is subject to copyright.All rights are reserved,whether the whole or part of the
materialisconcerned,specifically the rightsof translation,reprinting,reuseof illustrations,
recitation,broadcasting,reproductiononmicrofilmorinanyother way,andstorageindata
banks.Duplicationofthispublicationorpartsthereofispermittedonlyundertheprovisionsof
theGermanCopyrightLawofSeptember9,1965,initscurrentversion,andpermissionforuse
mustalwaysbeobtainedfromSpringer.ViolationsareliabletoprosecutionundertheGerman
CopyrightLaw.
The use of general descriptive names,registered names,trademarks,etc.in thispublication
doesnotimply,evenintheabsenceofaspecificstatement,thatsuchnamesareexemptfrom
therelevantprotectivelawsandregulationsandthereforefreeforgeneraluse.
Typeset&CoverDesign:ScientificPublishingServicesPvt.Ltd.,Chennai,India.
Printedinacid-freepaper
987654321
springer.com
Preface
Artificial evolutionary systems are computer systems, inspired by ideas from natural
evolution and related phenomena. The field has a long history, dating back to the
earliest days of computer science, but it has only become an established scientific and
engineering discipline since the 1990s, with packages for the commonest form, ge-
netic algorithms, now widely available.
Researchers in the Asia-Pacific region have participated strongly in the develop-
ment of evolutionary systems, with a particular emphasis on the evolution of intelli-
gent solutions to highly complex problems. The Asia-Pacific Symposia on Intelligent
and Evolutionary Systems have been an important contributor to this growth in im-
pact, since 1997 providing an annual forum for exchange and dissemination of ideas.
Participants come primarily from East Asia and the Western Pacific, but contributions
are welcomed from around the World.
This volume features a selection of fourteen of the best papers from recent APSIES.
They illustrate the breadth of research in the region, with applications ranging from busi-
ness to medicine, from network optimization to the promotion of innovation.
It opens with three papers in the general area of business and economics. Orito and
colleagues extend previous work on the application of evolutionary algorithms to
index fund optimization by incorporating local search in an unusual way: using the
genetic search to maximize the coefficient of determination between the fund’s return
rate and the market index (but not necessarily finding a linear relationship), and then
using local search to optimize the linearity. They demonstrate that this approach out-
performs direct search, yielding funds that perform substantially better as a surrogate
for the Tokyo Stock Price Index from 1997 to 2005.
Guo and Wong investigate the problem of learning Bayesian Networks from in-
complete data. They modify their group’s previous hybrid evolutionary algorithm for
learning from complete data. It uses essentially Friedman’s Structural Expectation
Maximization (SEM) algorithm as the outer loop, with a variant of their evolutionary
algorithm in the inner loop, replacing SEM’s hill-climbing phase. It differs from pre-
vious algorithms, which use the expected value to replace missing values, in using a
more sophisticated data completion process, which permits the use of decomposable
scoring metrics (specifically, information-based metrics) in the search process. They
use the algorithm in a direct-marketing application, demonstrating improved perform-
ance on that problem, though the technique would clearly extend to other domains –
DNA chip analysis, ecological data – where missing values cause serious difficulties.
VI Preface
Katai and his colleagues consider cooperative or 'local' currencies, and investigate
the design of such currencies to promote social and economic goals. They base their
analysis on fuzzy theory, and obtain interesting new results on the desirable operation
of such systems.
Networks have become a key area of complex systems research, with applications
ranging from communications to transport problems to the organisation of web pages.
The next six papers exemplify this trend, examining various aspects of network theory.
Leu and Namatame consider the problem of failure resilience in networks, such as
power distribution or communications networks. They apply evolutionary algorithms to
optimising the robustness of such networks to link failure, and are able to demonstrate
that, under certain circumstances, they are able to preserve important linkage properties
of the networks (notably, scale-freeness), while improving the failure resilience.
While Leu and Namatame consider robustness to link breakages in networks, Newth
and Ash consider instead robustness to disturbance, and the linearity of network
response. Again, they apply an evolutionary algorithm to optimise robustness. They
observe an interesting property, that the optimised networks they evolve exhibit hub-and-
star like topology, suggesting that this structure has inherent stability properties.
Komatsu and Namatame propose a heterogeneous flow control mechanism for pro-
tecting communications networks from attacks such as DDoS. They distinguish be-
tween altruistic protocols such as tcp, and uncontrolled protocols such as udp, using
open-loop congestion control mechanisms such as drop-tail for the former, and closed-
loop such as RED and CHOKe for the latter. Using simulations on a range of network
topologies, they demonstrate good performance in controlling excess traffic by com-
parison with homogeneous protocols, and propose extensions of this approach to
higher layers in the protocol stack.
Lin and Gen concentrate on the problem of network routing, specifically on finding
Shortest Path Routes (SPR) for Open Shortest Path First (OSPF) routing protocols.
They propose a new priority-based representation and genetic algorithm for this prob-
lem, and demonstrate its performance through a range of numerical experiments.
Network flow problems are a classic problem in the optimization literature; Gen,
Lin and Jo extend the usual problem, of maximizing network flow, into a bi-criteria
problem, maximizing network flow while minimizing network cost. They report on a
variant evolutionary multi-objective optimization algorithm incorporating Lamarckian
local search, and demonstrate its performance on a range of test problems.
A second paper from the same authors considers applications in logistics network
design, starting from the design of the network, and extending to vehicle routing and
automated vehicle dispatch. They introduce a priority-based Genetic Algorithm for the
task, applying variants to all three problems, with good results.
The final paper on network problems, by Lin and Gen, approaches the problem of
bi-criteria design of networks from a more general perspective. To illustrate their
approach, they tackle three separate design problems:
1. Shortest path, in which the conflicting objectives are to minimize transmission de-
lay while at the same time minimizing network cost
2. Spanning tree, in which the conflicting objectives are as above (i.e. minimizing
both transmission delay and network cost)
Preface VII
3. Network flow, in which the conflicting objectives are to maximize network flow
while at the same time minimizing network cost
The authors compare a number of representations and algorithms for these prob-
lems, generating interesting results showing that complex versions of these problems
can realistically be solved with today’s algorithms.
Sawazumi et al. investigate mechanisms to promote human creativity, proposing a
method based on “serendipity cards”, cards containing detailed information about a
theme. In so doing, they introduce a number of ideas and contexts from the Japanese
literature on idea generation not well known outside of Japan.
Cornforth et al tackle an important medical problem, that of recognition of medical
problems from imagery. Specifically, they concentrate on the issue of medical image
segmentation, in the context of assessment of retinopathy due to diabetes. They com-
bined wavelet data extraction methods with Gaussian mixture Bayesian classifiers,
generating substantially improvements over simpler methods, though not quite match-
ing expert-level human performance.
Gen et al tackle another highly practical problem, the problem of job-shop schedul-
ing in a shop where some machines may substitute for others for particular operations
(in the classical job-shop scheduling problem, each operation can be performed on
precisely one machine). They introduce a new multi-stage genetic algorithm, compar-
ing it with the state of the art in the field. They demonstrate very substantially im-
proved performance over a classical genetic algorithm, and GA augmented with a
form of local search, especially on hard problems. They demonstrate some improve-
ment in comparison with a particle-swarm/simulated annealing hybrid method, though
the differences are small.
Wong and Wong round out the volume with a paper showing that impressive speed
of evolutionary algorithms may be obtained at relatively low cost, through implemen-
tation on graphics processing units. They obtain very impressive performance indeed
on a range of benchmark optimization problems, especially for large population sizes.
Overall, the papers represent just a sample of the wide range of research in intelli-
gent and evolutionary systems being conducted in the Asia- Pacific region.. The grow-
ing maturity of its research culture portends an increasing contribution to international
research across the range of the sciences, and in intelligent systems in particular. We
hope this volume can serve as a stepping stone in this process, introducing some of the
work to a wider audience, and at the same time increasing international awareness of
one of this Asia-Pacific forum.

November 2008 Mitsuo Gen
David Green
Osamu Katai
Bob McKay
Akira Namatame
Ruhul Sarker
Byoung-Tak Zhang
Contents
Index Fund Optimization Using Genetic Algorithm and
Scatter Diagram Based on Coefficients of Determination
Yukiko Orito, Manabu Takeda, Hisashi Yamamoto .................... 1
Mining Bayesian Networks from Direct Marketing Databases
with Missing Values
Yuan Yuan Guo, Man Leung Wong ................................. 13
Fuzzy Local Currency Based on Social Network Analysis for
Promoting Community Businesses
Osamu Katai, Hiroshi Kawakami, Takayuki Shiose .................... 37
Evolving Failure Resilience in Scale-Free Networks
George Leu, Akira Namatame ...................................... 49
Evolving Networks with Enhanced Linear Stability Properties
David Newth, Jeff Ash............................................. 61
Effectiveness of Close-Loop Congestion Controls for DDoS
Attacks
Takanori Komatsu, Akira Namatame ................................ 79
Priority-Based Genetic Algorithm for Shortest Path Routing
Problem in OSPF
Lin Lin, Mitsuo Gen .............................................. 91
Evolutionary Network Design by Multiobjective Hybrid
Genetic Algorithm
Mitsuo Gen, Lin Lin, Jung-Bok Jo.................................. 105
Hybrid Genetic Algorithm for Designing Logistics Network,
VRP and AGV Problems
Mitsuo Gen, Lin Lin, Jung-Bok Jo.................................. 123
X Contents
Multiobjective Genetic Algorithm for Bicriteria Network
Design Problems
Lin Lin, Mitsuo Gen .............................................. 141
Use of Serendipity Power for Discoveries and Inventions
Shigekazu Sawaizumi, Osamu Katai, Hiroshi Kawakami,
Takayuki Shiose .................................................. 163
Evolution of Retinal Blood Vessel Segmentation Methodology
Using Wavelet Transforms for Assessment of Diabetic
Retinopathy
D.J.Cornforth, H.F.Jelinek,M.J.Cree,J.J.G.Leandro,J.V.B. Soares,
R.M. Cesar Jr. ................................................... 171
Multistage-Based Genetic Algorithm for Flexible Job-Shop
Scheduling Problem
Mitsuo Gen, Jie Gao, Lin Lin ...................................... 183
Implementation of Parallel Genetic Algorithms on Graphics
Processing Units
Man Leung Wong, Tien Tsin Wong ................................. 197
Author Index................................................... 217
Index Fund Optimization Using Genetic
Algorithm and Scatter Diagram Based on
Coefficients of Determination
Yukiko Orito1, Manabu Takeda2, and Hisashi Yamamoto2
1 Ashikaga Instituteof Technology
268-1, Ohmae-cho, Ashikaga, Tochigi 326-8558, Japan
[email protected]
2 TokyoMetropolitan University
6-6, Asahigaoka, Hino, Tokyo 191-0065, Japan
[email protected], [email protected]
Index fund optimization is one of portfolio optimizations and can be viewed as
a combinatorial optimization for portfolio managements. It is well known that
an index fund consisting of stocks of listed companies on a stock marketis very
useful for hedge trading if the total return rate of a fund follows a similar path
totherateofchangeofamarketindex.Inthispaper,weproposeamethodthat
consistsofageneticalgorithmandaheuristiclocalsearchonscatterdiagramsto
makelinearassociationbetweenthe returnratesandthe ratesofchangestrong.
A coefficient of determination is adopted as a linear associationmeasure of how
the return rates follow the rates of change. We then apply the method to the
Tokyo Stock Exchange. The results show that the method is effective for the
index fund optimization.
Keywords: Index Fund Optimization; Coefficient of Determination; Genetic Algo-
rithm; Heuristic Local Search.
1 Introduction
Index fund optimization is one of portfolio optimizations and can be viewed as
a combinatorialoptimizationfor portfoliomanagements.It is wellknownthata
group consisting of stocks of listed companies on a stock market is very useful
for hedge trading if the total return rates of a group follow a similar path to
the rates of change of a market index. Such a group is called an index fund. An
index fund has been used very extensively for the hedge trading, which is the
practice of offsetting the price risk on any cash market position by taking an
equal, but opposite position in a futures market [1]. In addition, there are some
studies report that the index funds have better performance than other mutual
funds [2, 3, 4].
The index fund optimization problem is one of NP-complete problems, and
it is impossible to solve it in reasonable time when the number of listed
M.Genetal.:IntelligentandEvolutionarySystems,SCI187,pp.1–11.
springerlink.com (cid:2)c Springer-VerlagBerlinHeidelberg2009