NAUTILUS Navigator: free search interactive multiobjective optimization without trading-off

摘要

We propose a novel combination of an interactive multiobjective navigation method and a trade-off free way of asking and presenting preference information. The NAUTILUS Navigator is a method that enables the decision maker (DM) to navigate in real time from an inferior solution to the most preferred solution by gaining in all objectives simultaneously as (s)he approaches the Pareto optimal front. This means that, while the DM reaches her/his most preferred solution, (s)he avoids anchoring around the starting solution and, at the same time, sees how the ranges of the reachable objective function values shrink without trading-off. The progress of the motion towards the Pareto optimal front is also shown and, thanks to the graphical user interface, this information is available in an understandable form. The DM provides preference information to direct the movement in terms of desirable aspiration levels for the objective functions, bounds that are not to be exceeded as well as the motion speed. At any time, (s)he can change the navigation direction and even go backwards if needed. One of the major advantages of this method is its applicability to any type of problem, as long as an approximation set of the Pareto optimal front is available and, particularly, to problems with time-consuming function evaluations. Its functionality is demonstrated with an example problem.