智能決策理論與方法ppt課件

1、)()(|)()(vFvbFaF)()(Re|)()()(RevApplevdvApplevCirclevd)()()(Re|)()(RevApplevBluevdvApplevd)()()(Re|)()()(RevApplevCirclevdvApplevCirclevdsxLxbvAauA)()(|)()(動物哺乳類鳥類企鵝食肉類蹄類飛禽類走禽類虎印度豹長頸鹿斑馬信天翁鷹駝鳥第1層第2層第3層第4層A0A1A1A2-11-11-110010110A0A1A1A2-11-11-110010110)log()log(),(NPNNPNNPPNPPNPIkjjjjjNPINPNPAE1),()(

2、0.9544)1610log(1610)166log(166),(NPI9056. 0)84log8484log84(168)86log8682log82(168)4 , 4(168)6 , 2(168)(0IIAE類似地，求出類似地，求出E(A1)，E(A2)，E(A3)。比較它們的大小，選擇。比較它們的大小，選擇期望信息熵最小的屬性作為根結(jié)點。依次構(gòu)造子決策樹，直至期望信息熵最小的屬性作為根結(jié)點。依次構(gòu)造子決策樹，直至一切的訓(xùn)練樣本均可以被正確分類。一切的訓(xùn)練樣本均可以被正確分類。)()()(ASplitiAGainAgainratiokjjjjjnNPnNPASpliti1)log()(

3、xxxxxf|)(xxxf11)(xxxeexforexf11)(11)(PppPpppEYYE1121)(21 1jijijiEttt012345678910-0.500.511.522.533.54012345678910-0.500.511.522.533.540510152025303540455010-310-210-110010110250 EpochsTraining-BluePerformance is 0.00225488, Goal is 0dxdyyxpyxffR),()()(liiiempyxflfR1)(1)()()(limfRfRempl)(min()(min(li

4、mfRfRempllhlhfRfRemp)4ln) 12(ln()()(2|wbwxdii1)(. .21minbwxytswwiiTdixxxxK)1 (),(21)/exp(),(2221ixxxxK)(tanh(),(2121kxxkxxKi),(,),(),(),(PcP210iiiikiiiiiiiiCciCccccccccc3kiki-ki-1 121ikieci1ikic2ic1ic0iics )(122)()()(11iiililssijisesmscmiikill1) 1)()()(),(/UCardXCCarddCkdUXNiiiidCdCN1),(),(s40.31s20

5、.49s10.14s30.061)() 1)(11irkjlijj) 1)() 1)(11iijjkilkjl)() 1)(11irkjlijjc1c2c3dc1c2c3d1911922.6134.527093.122612031.2138.720278.2116.1931515.519910.9228.718653.411112013.2217.110117.2218.812122.7113.615020.4123.226652.5114.8588.6113.514119.1120.46513.3123.321449.8140.417269.611c2c3c11c12c21c22c31c32c33cc1c2c