Constrained neural approaches to quadratic assignment problems

被引：22

作者：

Ishii, S

Sato, M

机构：

[1] Nara Inst Sci & Technol, Ikoma, Nara 6300101, Japan

[2] ATR, Human Informat Proc Res Labs, Seika, Kyoto 6190288, Japan

来源：

NEURAL NETWORKS | 1998年 / 11卷 / 06期

关键词：

combinatorial optimization; quadratic assignment problem; chaotic neural network; Potts spin; hard constraints;

D O I：

10.1016/S0893-6080(98)00077-X

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 discuss analog neural approaches to the quadratic assignment problem (QAP). These approaches employ a hard constraints scheme to restrict the domain space, and are able to obtain much improved solutions over conventional neural approaches. Since only a few strong heuristics for QAP have been known to date, our approaches are good alternatives, capable of obtaining fairly good solutions in a short period of time. Some of them can also be applied to large-scale problems, say of size N greater than or equal to 300. (C) 1998 Elsevier Science Ltd. All rights reserved.

引用

页码：1073 / 1082

页数：10

共 26 条

[1] CHAOTIC NEURAL NETWORKS [J].