A study on Understanding Effects on Genetic Operators with Visualization.

DOI Web Site

Bibliographic Information

Other Title
  • 可視化を用いた遺伝的演算効果把握に関する一考察
  • カシカ オ モチイタ イデンテキ エンザン コウカ ハアク ニ カンスル イチ コウサツ

Search this article

Abstract

Performance in searching solutions by GA depends on the genetic operators or their parameters such as the number of chromosomes, probability of crossover, that of mutation and so on. If the solution space is less or equal to 3 dimensions, we can display it in graphic and visually see the searching process. It is, however, difficult to visualize them in real combinational optimization problems for GA, since they have multidimensional searching space. Therefore, the efficient genetic operators and their parameters are usually determined by trial and error with actual applications. Moreover, this trial and error can give us limited information, just their performance. The purpose of this research is to grasp the searching process visually and get more information of not only the performance but also the effects of the genetic operations like the diversity of chromosomes in multidimensional problem. Then, we can feedback them into the genetic operations or their parameters. The proposed method employs Self Organizing Map(SOM) to this visualization. This paper shows that visualized data make us easier to know how the searching process is going on and find out more efficient genetic operators to keep the diversity comparing between them, which are difficult to get only from the comparison of fitness values.

Journal

Keywords

Details 詳細情報について

Report a problem

Back to top