Task-dependent evolution of modularity in neural networks

被引：13

作者：

Hüsken, M ^{[1
]}

Igel, C ^{[1
]}

Toussaint, M ^{[1
]}

机构：

[1] Ruhr Univ Bochum, Inst Neuroinformat, D-44780 Bochum, Germany

来源：

CONNECTION SCIENCE | 2002年 / 14卷 / 03期

关键词：

structure evolution; learning; measures for modularity; task-decomposition;

D O I：

10.1080/0954009021000047892

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

There exist many ideas and assumptions about the development and meaning of modularity in biological and technical neural systems. We empirically study the evolution of connectionist models in the context of modular problems. For this purpose, we define quantitative measures for the degree of modularity and monitor them during evolutionary processes under different constraints. It turns out that the modularity of the problem is reflected by the architecture of adapted systems, although learning can counterbalance some imperfection of the architecture. The demand for fast learning systems increases the selective pressure towards modularity.

引用

页码：219 / 229

页数：11

共 25 条

[1] [Anonymous], 1995, EVOLUTION OPTIMUM SE
[2] Bullinaria J, 2001, P 23 ANN C COGN SCI, P146
[3] DIFERDINANDO A, 2001, P 6 NEUR COMP PSYCH, P253
[4] Elman J.L., 1996, Rethinking Innateness: A Connectionist Perspective on Development, DOI DOI 10.1017/S0272263198333070
[5] Fodor J. A., 1983, MODULARITY MIND
[6] FRIEDRICH CM, 1996, 6 INT C INF PROC MAN, P951
[7] GRAU F, 1995, ADAPT BEHAV, V3, P151
[8] Optimization for problem classes -: Neural networks that learn to learn
Hüsken, M
Gayko, JE
Sendhoff, B
[J]. 2000 IEEE SYMPOSIUM ON COMBINATIONS OF EVOLUTIONARY COMPUTATION AND NEURAL NETWORKS, 2000, : 98 - 109
[9] HUSKEN M, 2001, LAT BREAK PAP GEN EV, P187
[10] Empirical evaluation of the improved Rprop learning algorithms
Igel, C
Hüsken, M
[J]. NEUROCOMPUTING, 2003, 50 : 105 - 123

← 1 2 3 →