Intelligent Strategies for Meta Multiple Criteria Decision Making

Intelligent Strategies for Meta Multiple Criteria Decision Making

by Thomas Hanne

Paperback(Softcover reprint of the original 1st ed. 2001)

Choose Expedited Shipping at checkout for guaranteed delivery by Wednesday, September 25


Multiple criteria decision-making research has developed rapidly and has become a main area of research for dealing with complex decision problems which require the consideration of multiple objectives or criteria. Over the past twenty years, numerous multiple criterion decision methods have been developed which are able to solve such problems. However, the selection of an appropriate method to solve a particular decision problem is today's problem for a decision support researcher and decision-maker.
Intelligent Strategies for Meta Multiple Criteria Decision-Making deals centrally with the problem of the numerous MCDM methods that can be applied to a decision problem. The book refers to this as a 'meta decision problem', and it is this problem that the book analyzes. The author provides two strategies to help the decision-makers select and design an appropriate approach to a complex decision problem. Either of these strategies can be designed into a decision support system itself. One strategy is to use machine learning to design an MCDM method. This is accomplished by applying intelligent techniques, namely neural networks as a structure for approximating functions and evolutionary algorithms as universal learning methods. The other strategy is based on solving the meta decision problem interactively by selecting or designing a method suitable to the specific problem, for example, the constructing of a method from building blocks. This strategy leads to a concept of MCDM networks. Examples of this approach for a decision support system explain the possibilities of applying the elaborated techniques and their mutual interplay. The techniques outlined in the book can be used by researchers, students, and industry practitioners to better model and select appropriate methods for solving complex, multi-objective decision problems.

Product Details

ISBN-13: 9781461356325
Publisher: Springer US
Publication date: 10/26/2012
Series: International Series in Operations Research & Management Science , #33
Edition description: Softcover reprint of the original 1st ed. 2001
Pages: 197
Product dimensions: 6.10(w) x 9.25(h) x 0.02(d)

Table of Contents

List of Figuresix
List of Tablesxi
1.MCDM problems1
2.Solutions of MCDM problems4
3.Decision processes and the application of MCDM methods5
4.Concepts of 'correct' decision making in MCDM methods9
5.Summary and conclusions14
2.The Meta Decision Problem in MCDM15
1.Methodological criticism in MCDM15
1.1Criticism on single concepts and methods15
1.2The discussion on the descriptive orientation of MCDM19
1.3Foundations by axioms of rational behavior22
2.The meta decision problem in MCDM24
2.1Formulation and foundation of the problem24
2.2Criteria for method selection25
2.2.1The suitability for a type of problem25
2.2.2Criteria based on solution concepts26
2.2.3Criteria oriented towards implementation28
2.2.4Criteria based on the specific decision situation30
2.3Scalar and multicriteria meta decision problems31
2.3.1Scalar evaluations of MCDM methods31
2.3.2Method choice as an MADM problem32
2.4The meta decision problem as a problem of method design34
2.4.1Determining the parameters of an MCDM method34
2.4.2Formalization of MCDM methods36
2.4.3A parameter optimization model37
2.5The problem of information acquisition39
2.5.1Implicit information40
2.5.2Explicit information41
3.Summary and conclusions44
3.Neural Networks and Evolutionary Learning for MCDM47
1.Neural networks and MCDM47
1.2The construction of neural networks working as traditional MCDM methods49
1.3Neural networks as an adaptive MCDM method54
2.Evolutionary learning55
2.1Evolutionary algorithms and neural networks56
2.2Evolutionary algorithms and MCDM59
3.Summary and conclusions61
4.On the Combination of MCDM Methods63
2.Properties of MCDM methods69
3.Properties of specific MCDM methods71
4.Properties of neurons and neural networks73
5.The combination of algorithms74
6.Neural MCDM networks75
7.Termination and runtime of the algorithm76
8.Summary and conclusions77
5.Loops--An Object Oriented DSS for Solving Meta Decision Problems79
1.Preliminary remarks79
2.Method integration, openness, and object oriented implementation80
3.A class concept for LOOPS84
4.Problem solving and learning from an object oriented point of view84
5.MADM methods in LOOPS87
6.Neural networks in LOOPS89
7.Neural MCDM networks in LOOPS90
8.Evolutionary algorithms in LOOPS91
9.An extended interactive framework95
10.Summary and conclusions98
6.Examples of the Application of LOOPS99
1.Some remarks on the application of LOOPS99
2.The learning of utility functions100
3.Stock selection106
4.Stock price prediction and the learning of time series113
5.Stock analysis and long-term prediction121
6.Method learning124
7.Meta learning127
8.An integrated proposal for the application of LOOPS131
9.Summary and conclusions132
7.Critical Resume and Outlook135
ASome basic concepts of MCDM theory163
2.Efficiency concepts and scalarizing theorems165
3.Utility concepts and other axiomatics166
BSome selected MCDM methods169
1.Simple additive weighting169
2.Achievement levels169
3.Reference point approaches170
4.The outranking method Promethee171
CNeural networks173
1.Introduction to neural networks173
2.Neural networks for intelligent decision support178
DEvolutionary algorithms181
1.Introduction to evolutionary algorithms181
2.The generalization of evolutionary algorithms186
EList of symbols189
FList of abbreviations193

Customer Reviews

Most Helpful Customer Reviews

See All Customer Reviews