Hybrid Approach to Multiattribute Decision Making under Uncertainty

The research described in this paper addresses issues of designing a computationally effective decision support system which can assist a decision maker in making an optimal choice between several discrete alternatives. A new hybrid approach to multi-attribute decision making under uncertainty incorporating Neural Networks and the Dempster-Shafer theory of evidence is introduced. A neural network is employed for representation and quantification of a decision maker’s pairwise preferences of one alternative over the other. The Dempster-Shafer theory of evidence is used for combining the evidences representing these preferences for modeling the choice of the most preferable alternative. The designed method also includes consideration of subjective judgments about attributes representing aggregated concepts along with quantitative attributes. The case study considered in the research has demonstrated the feasibility of the application of this approach to Fusion 2/3 Level problems, namely to threat assessment.