摘要
采用遗传算法(GA)作为归纳逻辑程序设计(ILP)的搜索策略，可以提高ILP方法的鲁棒性和适应性。文章简要叙述了对作者提出的遗传归纳逻辑程序设计(GILP) 算法作的改进，测试了选择策略对GILP算法收敛性能的影响。采用不同的选择策略不会影响算法的最终收敛结果，但会产生不同的选择压力，导致算法具有不同的收敛速率。
In order to enhance the robustness and adaptability of inductive logic programming(ILP) technique, genetic algorithm(GA) is used as the search strategy of ILP. After modifing the genetic inductive logic programming algorithm (GILP) proposed in this paper, an analysis of convergence performance of genetic inductive logic programming influnenced by using selection strategy is made. Experiment shows that different selection strategis don't have influence over the convergence result. However, different selection strategis bring out different selection pressure, which can lead to GILP with different convergence rates.
出处
《计算机工程》
CAS
CSCD
北大核心
2003年第5期8-9,15,共3页
Computer Engineering
基金
国家自然科学基金资助项目(60173014)
北京市自然科学基金资助项目(4022003)
