An improved immune genetic algorithm and its application in computer-aided landscape design

This paper proposes an improved immune genetic algorithm, and utilizes it in evaluating the results of computer- aided landscape design. After analyzing the related works and the flow chart of the standard immune genetic algorithm, an improved immune genetic algorithm is designed. The main modifications of our proposed immune genetic algorithm lie in the following aspects. 1) We modified the standard immune genetic algorithm using symbolic coding and full binary tree in the chromosomes to describe solutions. 2) The crossover operator with single point is used, and the cross point can be selected from the intermediate nodes and the root nodes. 3) The mutation operator is modified to avoid the dependence of mutation probability on the initial value. 4) The modified immune genetic algorithm not only can keep random global search ability, but also can avoid local premature convergence. Next, the landscape design evaluation results can be obtained by SVM, the parameters of which can be optimized by the proposed modified immune genetic algorithm. Finally, experiments are conducted on three datasets using an index system which is including 17 indexes. Experimental results demonstrate that the proposed scheme can effectively evaluate the quality of computer-aided landscape design. © Yao-Kuan and Yu-Hong.