抄録
Multi-objective optimization that can support agile and flexible decision-making has been highly required to deal with complex and global decision environment. We have proposed, in this paper, a general framework for solving multi-objective optimization problem relying on a prior articulation in tradeoff analysis among conflicting objectives. To overcome stiffness and shortcomings of the conventional methods, it has been developed by the virtue of simple subjective judgment and neural networks for identifying value function and its incorporation into an appropriate optimization method. Since the value function is decided by simple and relative responses that will be done separately from the searching process, DM's tradeoff analysis is not only very easy but also of small load compared with the conventional interactive methods. Another advantage is that we can carry cut MOP without particular knowledge about MOP and choose the most suitable optimization method. Such properties are very suitable to implement the algorithm on the Internet and to open it for the public users. A variety of illustrative applications are provided to verify effectiveness of the proposed method.
