摘要
检测器生成算法中采用随机搜索生成的检测器会产生大量重叠,而采用进化搜索收敛速度较慢.将两种搜索方式相结合,提出一种采用混合搜索的检测器生成算法,该算法将随机搜索产生的检测器集作为进化搜索的初始种群,使用遗传算法进化产生成熟检测器.使用二维人工数据测试算法.结果表明该算法能够以更少的检测器更精确地覆盖非自体空间,并能提升收敛速度.
In detector generation algorithm,random search method can generate a lot of overlap space among mature detectors,while evolutionary search method has relatively slow convergence speed.Combining these two search method,a detector generation algorithm based on mixed search is proposed.This algorithm generates a candidate set through random search firstly,and then uses this candidate set as the initial population of evolutionary search,at last evolves the mature detector set using genetic algorithm.The algorithm is tested using synthetic 2-D dataset.The experimental results demonstrate that mixed search method can achieved a more precise coverage of the nonself space with fewer detectors compared to random search,and has a higher convergence speed than evolutionary search.
出处
《小型微型计算机系统》
CSCD
北大核心
2010年第3期528-531,共4页
Journal of Chinese Computer Systems
基金
国家自然科学基金项目(60671049)资助
关键词
人工免疫
否定选择
进化搜索
随机搜索
收敛
artificial immune
negative selection
evolutionary search
random search
convergence