资源描述:
《边界条件熵的属性约简及在定性仿真中的应用_英文_》由会员上传分享,免费在线阅读,更多相关内容在教育资源-天天文库。
1、第38卷第10期中国科学技术大学学报Vol.38,No.102008年10月JOURNALOFUNIVERSITYOFSCIENCEANDTECHNOLOGYOFCHINAOct.2008ArticleID:025322778(2008)1021202209Featurereductionbasedonboundaryconditionalentropyanditsapplicationinqualitativesimulation1,222CHENGYu2sheng,ZHANGYou2sheng,HUXue2gang(1.SchoolofComputer
2、Science,AnqingTeachersCollege,Anqing246011,China;2.SchoolofComputerScience,HefeiUniversityofTechnology,Hefei230009,China)Abstract:SomenewdefinitionsofknowledgeroughentropyandboundaryconditionalentropyweregivenfromtheaspectofPawlaktopology.Thesedefinitionsaccuratelyreflectanideatha
3、tsetuncertaintycanbedescribedbyboundaryregion,whichwillmeasureknowledgeandroughsetuncertaintymoreaccurately.Meanwhile,animportantconclusionwasthatboundaryconditionalentropyofknowledgemonotonouslyreducedwiththediminishingofinformationgranularity.Throughanexampleofspringqualitatives
4、imulationreasoningtechnologycombinedwithknowledgeinformationentropybasedonroughsetstheory,aheuristicalgorithmforfeaturereductionwasproposedthatcouldbeusedtoeliminatetheredundancyinthequalitativedescriptionandthequalitativedifferentialequationswereobtainedfromthespringphysicalsyste
5、m.Experimentalresultsshowthattheroughsetstheoryisofgoodreliabilityandprospectinqualitativereasoningandsimulationandthatouralgorithmisaneffectivemethod.Keywords:qualitativesimulation;roughsetstheory;featurereduction;boundaryconditionalentropyCLCnumber:TP18Documentcode:A边界条件熵的属性约简及在
6、定性仿真中的应用1,222程玉胜,张佑生,胡学钢(1.安庆师范学院计算机与信息学院,安徽安庆246011;2.合肥工业大学计算机与信息学院,安徽合肥230009)摘要:从Pawlak拓扑的角度,给出了一种知识边界粗糙熵和边界条件熵的新定义,并反映出集合的不确定性可以通过边界域来描述的思想,证明了边界条件熵随着信息粒度的变小而单调减少的重要结论.弹簧定性仿真实例,结合定性推理技术,以边界条件熵为基础构造属性约简的启发式算法,消去定性描述中的冗余,获得了弹簧系统定性微分方程式.实验结果表明,粗集理论在定性推理与定性仿真技术中的重要应用价值,基于边界条件熵的属性
7、约简是有效的.关键词:定性仿真;粗集理论;属性约简;边界条件熵Received:2006210223;Revised:2007202201Foundationitem:SuportedbyNatureScienceFourdationofAnhuiProvince(070412061).Biography:CHENGYu2sheng(correspondingauthor),male,bornin1969,Ph.D/assistantprofessor.Researchfield:Roughsettheorydatamining.E2mail:chengy
8、saq@163.com第10期Featurereductionba