ON BETTER GENERALIZATION BY COMBINING 2 OR MORE MODELS - A QUANTITATIVE STRUCTURE-ACTIVITY RELATIONSHIP EXAMPLE USING NEURAL NETWORKS

被引:19
作者
AJAY
机构
[1] Department of Pharmaceutical Chemistry, University of California San Francisco, San Francisco, CA 94143, Box 0446
关键词
D O I
10.1016/0169-7439(94)00027-1
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
An important problem in statistical modeling (regression or classification) is to build models that generalize well. We propose a method to combine two or more models each obtained using cross-validation (CV), to generate a model that is better at generalization compared to the original ones. Such combinations are especially useful when cross-validation experiments indicate many possible solutions. The basic idea of the procedure described comes from actively making use of the predictions of cross-validation experiments, not just for calculating the CV error. The method is applied and is shown to perform well when building quantitative structure-activity relationships (QSARs) for steroids using neural networks. The proposed method has a much broader applicability than QSAR or neural networks.
引用
收藏
页码:19 / 30
页数:12
相关论文
共 9 条
[1]   A UNIFIED FRAMEWORK FOR USING NEURAL NETWORKS TO BUILD QSARS [J].
AJAY .
JOURNAL OF MEDICINAL CHEMISTRY, 1993, 36 (23) :3565-3571
[2]  
Breiman L., 1984, CLASSIFICATION REGRE
[3]  
Fedorov VV., 1972, THEORY OPTIMAL EXPT
[4]  
HERZBERG AM, 1986, UTILITAS MATHEMATICA, V29, P209
[5]   QUANTITATIVE RELATIONSHIPS BETWEEN STEROID STRUCTURE AND BINDING TO PUTATIVE PROGESTERONE RECEPTORS [J].
LEE, DL ;
KOLLMAN, PA ;
MARSH, FJ ;
WOLFF, ME .
JOURNAL OF MEDICINAL CHEMISTRY, 1977, 20 (09) :1139-1146
[6]  
Press W., 1986, NUMERICAL RECIPES FO
[7]  
ROGERS D, 1993, APPLICATION GENETIC
[8]   CROSS-VALIDATORY CHOICE AND ASSESSMENT OF STATISTICAL PREDICTIONS [J].
STONE, M .
JOURNAL OF THE ROYAL STATISTICAL SOCIETY SERIES B-STATISTICAL METHODOLOGY, 1974, 36 (02) :111-147
[9]   MODEL SELECTION VIA MULTIFOLD CROSS-VALIDATION [J].
ZHANG, P .
ANNALS OF STATISTICS, 1993, 21 (01) :299-313