摘要
构造性机器学习在构造覆盖领域时覆盖中心点的选择对覆盖领域的个数有直1接影响。针对这一问题,应用领域搜索算法,提出了一种新的构造性学习方法。把某个中心点和半径作为初始解,在邻近解中迭代,使覆盖网络逐步优化,直至不能再优为止。实验结果表明基于领域搜索的构造性学习算法可以使覆盖个数得到明显减少,不但可行而且行之有效。
When we structure coverage areas applying constructive machine learning algorithm, the choice of the mid-points of cover- age areas has a direct impact on the number of the areas. To solve this problem, a new constructive machine learning algorithm is proposed with the use of local search algorithm. A mid-point and radius of a coverage area can be seen a initial solution herative method can be used in the solution neighborhood to gradually optimize the covering networks, until the network can not be optimized. Experiments show that the number of coverage is decreased greatly using constructive machine learning algorithm based on local search and the algorithm is feasible and effective.
出处
《微计算机信息》
2012年第10期486-487,490,共3页
Control & Automation
基金
主持人:李萍
基金颁发部门:阜阳师范学院
项目名称:半监督学习的覆盖网络(2012FSKJ11)
主持人:程向阳
基金颁发部门:安徽省2010年高校省级教学质量与教学改革工程重点
项目名称:基于马氏链模型的教学发展性评价(20101984)
关键词
领域搜索
构造性学习
覆盖算法
Local search
Constructive machine learning
Covering algorithm