一种新的基于RS和NN的混合数据挖掘算法_英文_.pdfVIP

25 2 : V o l 25 No 2 2007 6 J o ur nal o f Gu angx i No rm al U niv er sity: N atural Science Edition J un 2007 A Nov el Dat a M ining A lg orit hm Based on RS and N N LUO Fei ( Scho ol of Elect ric and Inf or m atio n Eng ineer ing, Hunan U niver sity of T echno lo gy , Zhuzhou 412008, China) Abstract : A cco rding to t he adv antages and the pro blem s ex isting in r oug h set s theor y and neural netw or k of data m ining , an algo rithm b ased o n t he co mbinatio n o f ro ugh set s theor y and BP neura l netw or k a re present - ed T his alg or ithm dedu ct s data fr om data w areho use by u sing r ough set s deduct functio n, and then tran sf er s the dedu cted data t o the BP neur al netw or k as t raining data By data dedu ct , the expr ession o f t raining w ill be- com e clear er, and the scale of neur al net w o rk can be simplif ied A t the sam e t im e, neur al netw or k can solve r ough set s pr oblem of noise sen sitivit y T his p ap er also present s a co st function to ex press the relatio n ship betw een the amount o f tr aining dat a and t he pr ecisio n of neural netw or k , and to supply the standard f or the ch ange fr om r oug h set deduct t o neur al netw o rk t raining Key words : da ta m ining ; r oug h set s; neur al netw or k CLC number: T P 18 1Document code: A Article ID: 1001-6600 ( 2007) 02 -00 34 -04