A comparison of experimental designs in the development of a neural network simulation metamodel

被引:79
作者
Alam, FM [1 ]
McNaught, KR [1 ]
Ringrose, TJ [1 ]
机构
[1] Cranfield Univ, Dept Syst Engn, Appl Math & OR Grp, RMCS Shrivenham, Swindon SN6 8LA, Wilts, England
关键词
Latin Hypercube design; artificial neural network; metamodel; statistical experimental design;
D O I
10.1016/j.simpat.2003.10.006
中图分类号
TP39 [计算机的应用];
学科分类号
081203 [计算机应用技术]; 0835 [软件工程];
摘要
In this case study, we investigate the effects of experimental design on the development of artificial neural networks as simulation metamodels. A simple, deterministic combat model developed within the paradigm of system dynamics provides the underlying simulation. The neural network metamodels are developed using six different experimental design approaches. These include a traditional full factorial design, a random sampling design, a central composite design, a modified Latin Hypercube design and designs supplemented with domain knowledge. The results from this case study show how much impact the experimental design chosen for the neural network training set can have on the predictive accuracy, achieved by the metamodel. We compare the networks in terms of various performance measures. The neural network developed from the modified Latin Hypercube design supplemented with domain knowledge produces the best performance, outperforming networks developed from other designs of the same size. (C) 2004 Elsevier B.V. All rights reserved.
引用
收藏
页码:559 / 578
页数:20
相关论文
共 31 条
[1]
Response surface methodology: A neural network approach [J].
Anjum, MF ;
Tasadduq, I ;
AlSultan, K .
EUROPEAN JOURNAL OF OPERATIONAL RESEARCH, 1997, 101 (01) :65-73
[2]
Neural network as a simulation metamodel in economic analysis of risky projects [J].
Badiru, AB ;
Sieger, DB .
EUROPEAN JOURNAL OF OPERATIONAL RESEARCH, 1998, 105 (01) :130-142
[3]
BADIRU AB, 1993, NEURAL NETWORK SIMUL
[4]
Box G, 1987, EMPIRICAL MODEL BUIL
[5]
Chryssolouris G., 1990, Manufacturing Review, V3, P187
[6]
CHRYSSOLOURIS G, 1990, J MANUF SYST, V10, P166, DOI DOI 10.1016/0278-6125(91)90018-W
[7]
Coyle RG., 1996, SYSTEM DYNAMICS MODE
[8]
BAYESIAN PREDICTION OF DETERMINISTIC FUNCTIONS, WITH APPLICATIONS TO THE DESIGN AND ANALYSIS OF COMPUTER EXPERIMENTS [J].
CURRIN, C ;
MITCHELL, T ;
MORRIS, M ;
YLVISAKER, D .
JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION, 1991, 86 (416) :953-963
[9]
NEURAL NETWORK MODELS IN SIMULATION - A COMPARISON WITH TRADITIONAL MODELING APPROACHES [J].
FISHWICK, PA .
1989 WINTER SIMULATION CONFERENCE PROCEEDINGS, 1989, :702-710
[10]
Haykin S., 1998, NEURAL NETWORKS COMP