Title :
Function Finding and Constants Creation Method in Evolutionary Algorithm Based on Overlapped Gene Expression
Author :
Peng, Jing ; Tang, Chang-jie ; Yang, Dong-Qing ; Qiao, Shao-jie ; Zhang, Jing
Author_Institution :
Peking Univ., Beijing
Abstract :
Evolutionary algorithm based on overlapped gene expression (EAOGE) is a new technology of evolutionary algorithm which is inspired by the overlap gene expression in biological research. Different from existing works, EAOGE suggests a new expression structure of genes, and these genes have a probability to overlapped express in some segments. It uses chromosomes of fixed length to represent expression trees of different shapes and sizes. It does unconstrained search in the genome space and still ensures validity of the expression. This paper implements EAOGE algorithm and proposes a new constants creation method. Extensive experiments show that the method significantly improves the precision in the problems of function finding, and the precision of the new method is about 12.8 times to traditional algorithm at least.
Keywords :
data mining; genetic algorithms; biological research; constants creation method; evolutionary algorithm; function finding; overlapped gene expression; traditional algorithm; Biological cells; Biological information theory; Biology; Computer science; Data mining; Evolutionary computation; Gene expression; Genetic mutations; Genetic programming; Shape;
Conference_Titel :
Natural Computation, 2007. ICNC 2007. Third International Conference on
Conference_Location :
Haikou
Print_ISBN :
978-0-7695-2875-5
DOI :
10.1109/ICNC.2007.393