NEW METHOD OF MINING INCOMPLETE DATA

在线阅读 下载PDF 导出详情
摘要 Thedatausedintheprocessofknowledgediscoveryoftenincludesnoiseandincompleteinformation.Theboundariesofdifferentclassesofthesedataareblurandunobvious.Whenthesedataareclusteredorclassified,weoftengetthecoveringsinsteadofthepartitions,anditusuallymakesourinformationsysteminsecure.Inthispaper,optimalpartitioningofincompletedataisresearched.Firstly,therelationshipofsetcoverandsetpartitionisdiscussed,andthedistancebetweensetcoverandsetpartitionisdefined.Secondly,theoptimalpartitioningofgivencoverisresearchedbythecombingandpartingmethod,acquiringtheoptimalpartitionfromthreedifferentpartitionssetfamilyisdiscussed.Finally,thecorrespondingoptimalalgorithmisgiven.Therealwirelesssignalsofftencontainalotofnoise,andtherearemanyerrorsinboundarieswhenthesedataisclusteredbasedonthetradionalmethod.Inourexperimant,theproposedmethodimprovescorrectrategreatly,andtheexperimentalresultsdemonstratethemethod’svalidity.
机构地区 不详
出版日期 2013年04月14日(中国期刊网平台首次上网日期,不代表论文的发表时间)
  • 相关文献