Generalized Artificial Life Structure for Time-dependent Problems

作     者：TSAU Minhe KAO Weiwen CHANG Albert 

作者机构：Department of Mechanical Engineering Taiwan University of Science and Technology Department of Industrial Management Vanung University of Science and Technology 

出 版 物：《Chinese Journal of Mechanical Engineering》 (中国机械工程学报（英文版）)

年 卷 期：2009年第22卷第3期

页      面：317-324页

核心收录：

学科分类：0710[理学-生物学] 07[理学] 08[工学] 0836[工学-生物工程] 

主　　题：artificial life artificial intelligence generalized artificial life structure (GALS) 

摘      要：In recent years, more attention has been paid on artificial life researches. Artificial life（AL） is a research on regulating gene parameters of digital organisms under complicated problematic environments through natural selections and evolutions to achieve the final emergence of intelligence. Most recent studies focused on solving certain real problems by artificial life methods, yet without much address on the AL life basic mechanism. The real problems are often very complicated, and the proposed methods sometimes seem too simple to handle those problems. This study proposed a new approach in AL research, named ＂generalized artificial life structure（GALS）＂, in which the traditional ＂gene bits＂ in genetic algorithms is first replaced by ＂gene parameters＂, which could appear anywhere in GALS. A modeling procedure is taken to normalize the input data, and AL ＂tissue＂ is innovated to make AL more complex. GALS is anticipated to contribute significantly to the fitness of AL evolution. The formation of ＂tissue＂ begins with some different AL basic cells, and then tissue is produced by the casual selections of one or several of these cells. As a result, the gene parameters, represented by ＂tissues＂, could become highly diversified. This diversification should have obvious effects on improving gene fitness. This study took the innovative method of GALS in a stock forecasting problem under a carefully designed manipulating platform. And the researching results verify that the GALS is successful in improving the gene evolution fitness.