Evolution to the xtreme: Evolving evolutionary strategies using a meta-level approach

被引：8

作者：

Deugo, D ^{[1
]}

Ferguson, D ^{[1
]}

机构：

[1] Carleton Univ, Sch Comp Sci, Ottawa, ON K1S 5B6, Canada

来源：

CEC2004: PROCEEDINGS OF THE 2004 CONGRESS ON EVOLUTIONARY COMPUTATION, VOLS 1 AND 2 | 2004年

关键词：

D O I：

10.1109/CEC.2004.1330834

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

In this paper we describe a meta-level evolutionary system that uses a meta-level GA to evolve strategies that perform better than known good strategies on a test bed of mathematical optimization problems. We examine the effects of the meta-level components and parameters on the problem set in order to help others in choosing the components and parameters for their meta-GAs.

引用

页码：31 / 38

页数：8

共 18 条

[1]

[Anonymous], P 4 ANN C EV PROGR

[2]

[Anonymous], 1989, GENETIC ALGORITHM SE

[3]

BAECK T, 1993, P 5 INT C GEN ALG, P2

[4]

BAECK T, 1992, P 1 EUR C ART LIF CA, P227

[5]

DAVIS L, 1989, PROCEEDINGS OF THE THIRD INTERNATIONAL CONFERENCE ON GENETIC ALGORITHMS, P61

[6]

De Jong K. A., 1975, ANAL BEHAV CLASS GEN

[7]

FERGUSON D, 2002, THESIS CARLETON U

[8]

Freisleben B., 1993, Artificial Neural Nets and Genetic Algorithms. Proceedings of the International Conference, P392

[9]

FREISLEBEN B, 1993, P 9 INT C CONTR SYST, V2, P485

[10] OPTIMIZATION OF CONTROL PARAMETERS FOR GENETIC ALGORITHMS [J].

GREFENSTETTE, JJ .

IEEE TRANSACTIONS ON SYSTEMS MAN AND CYBERNETICS, 1986, 16 (01) :122-128

← 1 2 →