Handbook on Decision Making: Vol 1: Techniques and Applications

548 Pages·2010·10.86 MB·English

by  Chee Peng Lim

LakhmiC.JainandCheePengLim(Eds.)
HandbookonDecisionMaking:TechniquesandApplications
IntelligentSystemsReferenceLibrary,Volume4
Editors-in-Chief
Prof.JanuszKacprzyk Prof.LakhmiC.Jain
SystemsResearchInstitute UniversityofSouthAustralia
PolishAcademyofSciences Adelaide
ul.Newelska6 MawsonLakesCampus
01-447Warsaw SouthAustralia5095
Poland Australia
E-mail:[email protected] E-mail:[email protected]
Furthervolumesofthisseriescanbefoundonourhomepage:springer.com
Vol.1.ChristineL.MumfordandLakhmiC.Jain(Eds.)
ComputationalIntelligence:Collaboration,Fusion
andEmergence,2009
ISBN978-3-642-01798-8
Vol.2.YuehuiChenandAjithAbraham
Tree-StructureBasedHybrid
ComputationalIntelligence,2009
ISBN978-3-642-04738-1
Vol.3.AnthonyFinnandSteveScheding
DevelopmentsandChallengesfor
AutonomousUnmannedVehicles,2010
ISBN978-3-642-10703-0
Vol.4.LakhmiC.JainandCheePengLim(Eds.)
HandbookonDecisionMaking:Techniques
andApplications,2010
ISBN978-3-642-13638-2
Lakhmi C.Jain and Chee Peng Lim (Eds.)
Handbook on Decision Making
Vol 1: Techniques andApplications
123
Prof.LakhmiC.Jain
UniversityofSouthAustralia
SchoolofElectrical&
InformationEngineering
KESCentre
5095AdelaideSouthAustralia
MawsonLakesCampus
Australia
E-mail:[email protected]
Dr.CheePengLim
UniversityofScienceMalaysia
SchoolofElectricaland
ElectronicEngineering
EngineeringCampus
14300NibongTebal,Penang
Malaysia
E-mail:[email protected]
ISBN 978-3-642-13638-2 e-ISBN978-3-642-13639-9
DOI 10.1007/978-3-642-13639-9
Intelligent SystemsReference Library ISSN1868-4394
Library of Congress Control Number:2010928590
(cid:2)c 2010 Springer-VerlagBerlin Heidelberg
Thisworkissubjecttocopyright.Allrightsarereserved,whetherthewholeorpart
of the material is concerned, specifically therights of translation, reprinting,reuse
ofillustrations, recitation,broadcasting, reproductiononmicrofilm orinanyother
way, and storage in data banks. Duplication of this publication or parts thereof is
permittedonlyundertheprovisionsoftheGermanCopyrightLawofSeptember9,
1965, in its current version, and permission for use must always be obtained from
Springer. Violations are liable to prosecution undertheGerman Copyright Law.
The use of general descriptive names, registered names, trademarks, etc. in this
publication does not imply, even in the absence of a specific statement, that such
names are exempt from the relevant protective laws and regulations and therefore
free for general use.
Typeset&CoverDesign:ScientificPublishing ServicesPvt. Ltd., Chennai, India.
Printed on acid-free paper
9 8 7 6 5 4 3 2 1
springer.com
Preface
Decision making arises when we wish to select the best possible course of action
from a set of alternatives. With advancements of the digital technologies, it is
easy, and almost instantaneous, to gather a large volume of information and/or
data pertaining to a problem that we want to solve. For instance, the world-wide-
web is perhaps the primary source of information and/or data that we often turn to
when we face a decision making problem. However, the information and/or data
that we obtain from the real world often are complex, and comprise various kinds
of noise. Besides, real-world information and/or data often are incomplete and
ambiguous, owing to uncertainties of the environments. All these make decision
making a challenging task. To cope with the challenges of decision making, re-
searchers have designed and developed a variety of decision support systems to
provide assistance in human decision making processes.
The main aim of this book is to provide a small collection of techniques
stemmed from artificial intelligence, as well as other complementary methodolo-
gies, that are useful for the design and development of intelligent decision support
systems. Application examples of how these intelligent decision support systems
can be utilized to help tackle a variety of real-world problems in different do-
mains, e.g. business, management, manufacturing, transportation and food indus-
tries, and biomedicine, are also presented. A total of twenty chapters, which can
be broadly divided into two parts, i.e., (i) modelling and design techniques for
intelligent decision support systems; and (ii) reviews and applications of intelli-
gent decision support systems, are included in this book. A summary of each
chapter is as follows.

Part I Modelling and Design Techniques for Intelligent Decision Support
Systems

An overview of intelligent decision making is presented in Chapter 1. The general
aspects of decision making, decision quality, and types of decision support sys-
tems are explained. A number of intelligent techniques stemmed from artificial
intelligence that are useful for the design and development of intelligent decision
support systems are described. Application examples of these intelligent tech-
niques, as well as their hybrid models, are also highlighted.
In Chapter 2, an intelligent decision support systems engineering methodology
for designing and building intelligent decision support systems is described. The
proposed methodology comprises four phases: project initiation, system design,
system building and evaluation, and user’s definitive acceptance. The usefulness
of the proposed methodology is assessed in academic settings with realistic case
VI Preface
studies, and satisfactory results are reported. The implication of the proposed
methodology in providing a systematic software engineering oriented process for
users to develop intelligent decision making support systems is discussed.
In a highly competitive market, the design of products becomes a challenging
task owing to diversified customer needs and complexity of technologies. In
Chapter 3, a framework that describes the relationships of the product design
problems, product design processes, shape design processes, shape design meth-
ods and tools with consideration of the functional, ergonomic, emotional, and
manufacturing requirements, is presented. A decision support system to assist
designers in designing product shapes is developed. A case study on scooter
shape design is conducted. Applicability of the decision support system to plan-
ning the shape design process and creating a scooter shape following the planned
processes is demonstrated.
A Tree-based Neural Fuzzy Inference System (TNFIS) for model formulation
problems in time series forecasting, system identification, as well as classification
problems is suggested in Chapter 4. The proposed approach takes the imprecise
nature of decision makers' judgements on the different tacit models into considera-
tion. The learning algorithm of the TNFIS consists of two phases: Piaget's action-
based structural learning phase, and a parameter tuning phase. Knowledge in the
form of fuzzy rules is created using the TNFIS, and visualization techniques are
proposed so that the decision maker can better understand the formulated model.
The effectiveness of the TNFIS is demonstrated using benchmark problems.
A general approach to decision making in complex systems using agent-based
decision support systems is described in Chapter 5. The approach contributes to
decentralization and local decision making within a standard work flow. A layered
structure is adopted to address issues involving (i) data retrieval, fusion, and pre-
processing; (ii) data mining and evaluation; and (iii) decision making, alerting,
solutions and predictions generation. The agent-based decision support system is
applied to evaluate the impact of environmental parameters upon human health in
a Spanish region. It is found that the system is able to provide all the necessary
steps for decision making by using computational agents.
In Chapter 6, single-criterion and multiple-criteria decision analysis for sustain-
able rural energy policy and planning is described. A number of single-criterion
and multiple-criteria energy decision support systems are analysed, with their
strengths and limitations discussed. A sustainable rural energy decision support
system, which combines quantitative and qualitative criteria and enables the pri-
orities of a group of prospective users to be considered in decision analysis, is
described. Novelty of the decision support system lies in its ability to match rural
community’s needs in developing countries to appropriate energy technologies,
thereby improving livelihoods and sustainability. A study on using the decision
support system for energy analysis and planning of a remote community in Co-
lombia is presented.
It is argued that decision making is the bridge between sensation and action, i.e.
the bridge between processing of stimulus input and generation of motor output.
In Chapter 7, a decision making model known as the complementary decision
making system is proposed. The model is based complementary learning that
Preface VII
functionally models the lateral inhibition and segregation mechanisms observed in
the human decision making process, i.e., in prefrontal and parietal lobes neural
basis of decision making. Fuzzy rules are generated to inform users how confi-
dent the system is in its predictions. To assess the performance of the proposed
model, a number of benchmark medical data sets are used. The results compared
favourably with those from other machine learning methods.
A variety of forecasting techniques are used by a lot of major corporations to
predict the uncertain future in an attempt to make better decisions which affect the
future of the organizations. In Chapter 8, a forecasting support system based on
the exponential smoothing scheme for forecasting time-series data is presented.
Issues related to parameter estimation and model selection are discussed. A
Bayesian forecasting support system, i.e., SIOPRED-Bayes, is described. The
system incorporates existing univariate exponential smoothing models as well as
some generalizations of these models in order to deal with features arising in eco-
nomic and industrial scenes. Its application to water consumption forecasting is
demonstrated.
Partially observable Markov decision processes provide a useful mathematical
framework for agent planning under stochastic and partially observable environ-
ments. In Chapter 9, a memory-based reinforcement learning algorithm known as
reinforcement-based U-Tree is described. It is able to learn the state transitions
from experience and build the state model by itself based on raw sensor inputs.
Modifications to U-Tree’s state generation procedure to improve the effectiveness
of the state model are also proposed. Its performance is evaluated using a car-
driving task. In addition, a modification to the statistical test for reward estima-
tion is suggested, which allows the algorithm to be benchmarked against some
model-based approaches with well-known problems in partially observable
Markov decision processes.
The Fuzzy Inference System (FIS) has been demonstrated to be a useful model
in undertaking a variety of assessment and decision making problems. In Chapter
10, the importance of the monotonicity property of an FIS-based assessment
model is investigated. Specifically, the sufficient conditions for an FIS-based
assessment model to satisfy the monotonicity property are derived. In addition, a
Failure Mode and Effective Analysis (FMEA) framework with an FIS-based Risk
Priority Number (RPN) model is examined. A case study of the applicability of
the FMEA framework to a semiconductor manufacturing process is conducted.
The results obtained indicate the importance of the monotonicity property of the
FIS-based RPN model in tackling assessment and decision making problems.

Part II Reviews and Applications of Intelligent Decision Support Systems

A thorough study on the use of decision support systems in the transportation
industry is presented in Chapter 11. A taxonomy for classifying transportation
decision support systems is described. The usefulness of transportation decision
support systems in solving different types of decision problems is examined.
Methodologies of decision making as well as information technologies that are
useful for developing transportation decision support systems are also discussed.
VIII Preface
A useful review on a large variety of decision support systems that are applied to
different transportation sectors, which cover road, urban, air, rail, and seaborne
transportation, is included.
Application of decision support systems to the food industry, and in particular
the seafood industry, is presented in Chapter 12. The food industry is different to
many other industries, since the nature of the products and ingredients can change
dramatically with time. Traceability is important in order to know the history of
the product and/or ingredient of interest. By exploiting product traceability, the
flow of data can be used for decision support. It is pointed out that decision sup-
port systems are beneficial to a number of areas in food processing, which include
lowering environmental impact of food processing, safety management, process-
ing management, and stock management. The usefulness of decision support
systems for the meat industry, food producers, inventory management and replen-
ishment for retailers are also described.
Creative city design is a multi-facet problem which involves a wide range of
knowledge and a diverse database. Based on rough sets, a decision support sys-
tem that is able to help decision-makers leverage resources with information tech-
nology for creative city design is presented in Chapter 13. The design rules of
creative city development by urban design experts are also described. Rough set
theory is applied to select the decision rules and measure the current status of
Japanese cities. A prototype, i.e., Urban Innovators Systems, to demonstrate the
usefulness of the approach in building a collaborative model of creative city with
public participation is discussed.
A major aspect of decision making is in making buying and selling decisions.
In Chapter 14, a combinatorial auction mechanism where bidders can submit mul-
tiple prices (pessimistic, ideal, and optimistic values) in a single package is de-
scribed. A new operationalization on the auctioneer-bidder relationship based on
the type of bids or triangular possibility distribution is proposed. The proposed
approach is evaluated with test problems in a fuzzy auction environment. The
analysis reveals that the fuzzy solution interval provides both negotiation and risk
assessment capability for the auctioneer.
Accidental building fires cause many fatalities and property losses to the com-
munity. Artificial neural networks have been shown to be an efficient and effec-
tive decision making models in fire safety applications. In Chapter 15, a hybrid
neural network model that combines Fuzzy ART and the General Regression
Neural Network is proposed. A series of experiments using benchmark datasets to
examine the usefulness of the network in tackling general data regression prob-
lems is first conducted. A novel application of the proposed hybrid network to
predicting evacuation time during fire disasters is described. The results demon-
strate the efficacy of the proposed network in undertaking fire safety engineering
problems.
In Chapter 16, the use of the path-converged design, which is a nonparametric
approach, for decision making in optimal migration strategy in urban planning is
examined. A study to identify existing population agglomeration for small, me-
dium, and large cities from both regional and urban perspectives and to evaluate
the efficiency of existing population agglomeration in urban planning is first
Preface IX
conducted. Identification based on path-converged design reveals inefficiency in
existing population agglomeration in China. Based on the identified population
agglomeration and the inefficiency of agglomeration, a number of population migra-
tion decisions to eliminate inefficiency of population allocation are discussed.
In Chapter 17, the use of a number of fuzzy neural networks to superficial ther-
mal images against the true internal body temperature is described. Comparison
between global and local semantic memories as well as Mamdani and Takagi-
Sugeno-Kang model of fuzzy neural networks are presented. A series of experimen-
tal studies using real data from screening of potential SARS patients is conducted.
The experimental results of temperature classification based on thermal images are
analysed. A comparison between various global and local learning networks is pre-
sented. The outcomes demonstrate the potential of fuzzy neural networks as an intel-
ligent medical decision support tool for thermal analysis, with the capability of
yielding plausible semantic interpretation of the system prediction to domain users.
Electroencephalogram (EEG) is one of the most important sources of informa-
tion in therapy of epilepsy, and researchers have addressed the issue of engaging
decision support tools for such a data source. In Chapter 18, the application of a
novel fuzzy logic system implemented in the framework of a neural network for
classification of EEG signals is presented. The proposed network constructs its
initial rules by clustering while the final fuzzy rule base is determined by competi-
tive learning. Both error backpropagation and recursive least squares estimation
techniques are used for tuning premise and consequence parameters of the net-
work. Applicability of the network to EEG signal classification is demonstrated.
In Chapter 19, a case base reasoning system for differentiation based on altered
control of saccadic eye movements in Attention-Deficit Hyperactivity Disorder
(ADHD) subjects and a control group is described. The TA3 system, an intelli-
gent decision support system that incorporates case based reasoning into its
framework, is used to retrieve and apply previous ADHD diagnostic cases to
novel problems based on saccade performance data. The results demonstrate that
the proposed system is able to distinguish ADHD from normal control subjects,
based on saccade performance, with increasing accuracy.
In Chapter 20, the use of a brain inspired, cerebellar-based learning memory
model known as pseudo self-evolving cerebellar model articulation controller to
model autonomous decision-making processes in dynamic and complex environ-
ments is described. The model adopts an experience-driven memory management
scheme, which has been demonstrated to be more efficient in capturing the inher-
ent characteristics of the problem domain for effective decision making. Applica-
bility of the model is evaluated using dynamics of the metabolic insulin regulation
mechanism of a healthy person when perturbed by food intakes. The model is use-
ful for capturing complex interacting relationships of the blood glucose level, the
food intake, and the required blood insulin concentration for metabolic homeostasis.