Evaluation function optimization by the Demonstrational Interface using Distributed GA

Abstract

Generally, it is difficult to define evaluation function, since it requires the special knowledge over the problem. To solve this problem, we propose a method by using Distributed Genetic Algorithm (DGA), The proposed system shows samples generated from each subgroup of DGA and the evaluation function is automatically tuned by choosing the sample with a user's intention. Reflection of individual taste is also enabled simultaneously. In this research, the evaluation function optimization by the Demonstrational Interface was applied to Fuzzy graph automatic drawing, and the validity of the proposed method was analyzed.