HDN-GEP: A Novel Gene Expression Programming with High Density Node
Gene expression programming (GEP) is a new member of evolutionary computation family, and is successful in symbolic regression and function finding in the field of data mining. However, GEP is difficult to find power functions with high ranks. To tackle this problem, this study proposes a novel GEP algorithm named HDN GEP. The main contributions include: (1) a new structure named HDN (high density node) is proposed that makes each bit in chromosome express more genetic information, (2) a HDN-GEP algorithm is proposed to solve the high or super-high power polynomial function funding, (3) the efficiency of evolution and the ability of GEP in function finding is improved based HDN-GEP, and (4) extensive experiments demonstrate that HDN GEP algorithm can find high power functions with short chromosome, whereas it can not be solved efficient by traditional GEP.
Yu Chen Changjie Tang Chuan Li Yue Wang Ning Yang Mingfang Zhu
College of Computer Science, Sichuan University, Chengdu, China College of Computer Science, Sichuan University, Chengdu, China Department of Computer Science and T
国际会议
长沙
英文
60-64
2008-10-20（万方平台首次上网日期，不代表论文的发表时间）