Interactive Evolutionary Algorithms for Multi-Objective Optimization: Design and Validation of a Hybrid Interactive Reference Point Method

Many real world problems have conflicting multiple objectives which a decision maker would like to optimize simultaneously. Thus, the decision maker may obtain many optimal solutions from a solution method. Subsequently, the user has the difficult task to pick one solution.Interactive methods focus on solutions which are located in the users region of interest. The author, Madan Sathe, analyses and classifies essential existing interactive classical methods and provides a new grouping approach for existing interactive evolutionary methods. After a crossfertilization of these two fields a new Hybrid Interactive Reference Point Method is designed and implemented within a powerful graphical tool. The tool guides the user to new regions of interest step by step while yielding detailed information about solutions at each step.Finally, the new approach is validated on test and real world problems. Currently, the scope of the tool is limited to non-linear optimization problems.The book is aimed at people who are interested in the field of multi-objective optimization and who would like to gain information about interactive methods and the realization in graphical tools.

Madan Sathe, Dipl. Inf.: Studies of Computer Science at the Technical University of Dortmund, Germany. PhD Student at the Department of Computer Science at the University of Basel, Switzerland.