A Multidimensional Search Space Using Interactive Genetic Algorithm

This paper applied an Interactive Genetic Algorithm (IGA) technique to design an visualization environment for search space. In past IGA works with a small size of population and generations to prevent the user fatigue, in this way small search space is used for evolving solutions. The proposed work is based on the use of objective function for assigning fitness, visualization of multidimensional search space and interaction of user after several generations. The interaction of user only required for proposing genes values at different locations of particular generation. In our proposed method, the multidimensional data of each generation is represented as 2-D graphs. The selection and interaction of interesting part of search space helps to generate more potential solutions. For experiment, we have selected Parametric L-System, in which both symbols and numerical parameters are evolved using Genetic Algorithm (GA). The output of this visualization helps to generate random strings and parameters, which are used to construct rules for Parametric L-System. Since Parametric L-System is involved, 2 Layered GA is introduced in which upper layer evolved symbols and lower layer evolved parameters.