The Hybridized Optimization with Gene Expression Programming and Niche Technology for Association Rule Mining
Gene Expression Programming (GEP) has found its wide application in the problem of data mining, cellular automata rules for the density-classification and so on. However, unfortunately, the research has shown that the standard GEP, performing worse especially in multi-level spatial search, has certain weaknesses, such as, slow convergence and low accuracy of solutions. Corresponding to this question, a novel Niche GEP (NGEP) is firstly presented in this paper. The algorithm is proposed to establish multi-population and storage structure among the initial chromosomes; the decoded chromosomes and fitness are modified and implemented. And then we apply NGEP to the field of mining association rules. The experimental results show that our algorithm performs better than the alternative evolutionary algorithm in terms of diversity of population and precision; besides, it can discover more association rules that cannot be extracted by other similar method.
Association rules Gene Expression Programming Niche Genetic Algorithm Data Mining
Jie Yang Yunliang Chen Dehua Li Qing Chen Lei Chen Gang Huang
the Institute of Pattern Recognition and Artificial Intelligence, State Commission Research Laborato Dept.Of Computer Science and Technology, China University of Geosciences Jinshan Campus of No.3 Middle School, Fuzhou
国际会议
武汉
英文
2007-09-21（万方平台首次上网日期，不代表论文的发表时间）