摘要
分类器集成是一种解决复杂模式识别问题的有效方法,它通过形成一组分类器并将它们的结果进行组合,可以显著地提高分类系统的泛化推广能力。论文将分类器集成技术用于核爆地震信号的模式识别,并以前馈神经网络作基分类器为例,研究了不同的分类器个体生成方法和决策形成规则对识别效果的影响。
As a powerful solution to difficul t pattern recognition problem,classifier ensemble can significantly improve the e xtensibility of classifying systems through training a finite number of classif iers and combining their results.In this paper,classifier ensemble is used to classify nuclear explosion and nature earthquake by the seismic recording.Take the forward-feed neural network as an example,the generation methods of differ ent classifier and the influences of the deci-sion formation rule on the recog nition effect are analyzed.
引文
1.吴忠良.核爆炸地震学概要[M].北京:地震出版社,1994
2.CHChen,ICLin.PatternanalysisandclassificationwiththenewACDAseismicsignaturedatabase[R].ReportAD-A015,925,ArmsControlandDisarmamentAssociation,Washington,DC,1975
3.DaizhiLiuetal.Fractalanalysiswithapplicationstoseismologicalpatternrecognitionofundergroundnuclearexplosion[J].SignalPro-cessing,2000;80:1849~1861
4.刘代志等.分形分析与核爆地震模式识别[J].模式识别与人工智能,1997;10(2):153~158
5.赵克等.核爆地震波的时频特征[J].核电子学与探测技术,2000;20(4):272~278
6.赵克,刘代志.时频表示的奇异值特征[J].第二炮兵工程学院学报,1999;13(4):42~47
7.赵克等.目标识别的特征量约束[J].信号处理,1999;15(增刊):53~456
8.李夕海等.目标识别中的特征相空间吸引子分析[J].信号处理,2001;17(增刊)
9.王飞等.核爆地震的统计特征分析与识别研究[C].见:陕西地球物理文集,西安:西安地图出版社,1999:22~25
10.刘代志等.吸引子分析与核爆地震模式识别[J].电子学报,1997;25(10):122~125
11.刘代志等.FUZZY综合评判与模式识别[J].第二炮兵工程学院学报,1998;12(1):16~19
12.苏娟等.一种改进的K近邻法[J].第二炮兵工程学院学报,1997;11(4):16~19
13.ZhaoKeetal.Anewalgorithmforlineardiscriminantfunctionanditsapplicationstopatternrecognitionofnuclearexplosion[C].In:FourthInternationalConferenceonSignalProcessingProceedings,October12~16,1998,Beijing,China,PublishingHouseofElectronicsIndustry,1998:1237~1240
14.刘代志等.地下核爆炸的自动识别研究[J].核电子学与探测技术,1998;18(1):12~16
15.HansenLK,SalamonP.Neuralnetworkensembles[J].IEEETransPatternAnalysisandMachineIntelligence,1990;12(10):993~1001
16.SchapireRE.Thestrengthofweaklearnability[J].MachineLearning,1990;5(2):197~227
17.BreimanL.Baggingpredictors[J].MachineLearning,1996;24(2):123~140
18.FreundY,SchapireRE.Adecision-theoreticgeneralizationofon-linelearningandanapplicationtoboosting[J].JournalofComputerandSystemSciences,1997;55(1):119~139
19.张立明.人工神经网络的模型及其应用[M].上海:复旦大学出版社,1992