基于代数和几何组合相似度的区间值系统不确定性度量

打开文本图片集

中图分类号：TP181 文献标志码：A

'，²（1，安徽合肥230011；2，湖北十堰442002)

An uncertainty measurement for interval valued systems based on algebra geometric combination similarity

', ²

(1 ，Anhui VocationalTechnical,Hefei，Anhui230,Cina; 2School InteligentConnected Vehicles， ，Shiyan,442002,China)

Abstract：Uncertaintymeasurement，asanimportantdataevauationtoolinthefieldsmachinelearninggranularcomputing，can quantifytheuncertaintydependenciesbetweendataatributes.However，existing intervalvalueinformationsystemuncertanty measurementmethodsdonotcosiderthegeometricstructurebetweendataatrbutes，whichafectstheaccuracyuncertaintymeasurementresults.Toimprovethisissue，firstly，ahybridsimilaritymethodcombiningalgebraicgeometricperspectivesisproposed forintervalvaluedinformationsytems；hen,asedonthenewsiilarityewintervalvaluedroughsetmodelwasconstructednd thetheoryinformationgraulationinformationstructureforintervalvaluedinformationsystemssproposed；Finalyfouruncertaintymeasurementmethodsforintervalvaluedinformationsystemsweredefined，namelyknowledgegranularity，inforationamount，roughentropy，informationentropy.Theoreticalevidencehasshowntheefectivenessthesefourmethodsinucertainty measurement.Thenumericalexperimentalresultsonthefacerecognitiondatasetshowthatthefourproposeduncertaintymeasurement methodsallhavegoodmeasurementperformance，comparedwithexistinguncertaintymeasurementmethods，thefourproposed methods have higher measurement performance.

Keywords；roughset;intervalvalueinformationsstem;uncertaintymeasurement;knowledgeganularity;informationamount;ough entropy;information entropy

粗糙集理论是一种处理数据分析中不精确性和不确定性的数学工具,已成功应用于智能系统、机器学习、决策分析和模式识别等领域中[1-3]。（剩余36289字）