边界条件熵的属性约简和在定性仿真中的应用_英文_.pdf

第38卷第10期 中国科学技术大学学报 Vol . 38 ,No . 10 2 0 0 8 年 1 0 月 J OURNAL OF UNIVERSITY OF S CIEN CE AND TECHN OL O G Y OF CHINA Oct . 2 0 0 8 ( ) Article ID 2008 Feature reduction based on boundary conditional entropy and its appl ication in qual itative simulation C H EN G Yusheng1 ,2 ,Z HAN G Yo usheng2 , H U Xuegang2 ( 1. S chool of Comp uter S cience , A nqi ng Teachers Col leg e , A nqi ng 2460 11 , Chi na; ) 2 . S chool of Comp uter S cience , H ef ei Uni vers ity of Technology , H ef ei 230009 , Chi na Abstract : So me new definitio n s of knowledge rough ent rop y and boundary conditional ent rop y were given f ro m t he a sp ect of Pawlak top olo gy . The se definitio n s accurat ely reflect an idea t hat set uncert aint y can be de scribed by bo undary region , w hich will mea sure knowledge and rough set uncert aint y more accurat ely . Meanw hile , an import ant conclu sion wa s t hat boundary conditional ent rop y of knowledge monotonou sly reduced wit h t he dimini shin g of infor mation granularit y . Thro ugh an examp le of sp ring qualit ative simulation rea soning t echnolo gy co mbined wit h knowledge infor mation ent rop y ba sed on rough set s t heory , a heuri stic algorit hm for f eat ure reduction wa s p ropo sed t hat coul d be u sed to eliminat e t he redundancy in t he qualit ative de scrip tion and t he qualit ative diff erential equatio n s were obt ained f ro m t he sp rin g p hy sical