| ²é¿´: 675 | »Ø¸´: 1 | ||
stonechao1½ð³æ (ÖøÃûдÊÖ)
|
[ÇóÖú]
¹ØÓÚSVMÀ¸çÀÊÈÕ³Ë×Ó¶ÔÓ¦µÄµÄÖ§³ÖÏòÁ¿ÎÊÌâ?ÓÃcvx_boxÉè¼ÆSVM
|
|
load FV600 % FV600ÊÇÒ»¸öÊý¾Ý¼¯£¬X1¶ÔÓ¦ÌØÕ÷ÏòÁ¿£¬y¶ÔÓ¦±êÇ©£¬ÊǸöÁ½ÀàÎÊÌâ %ÏÂÃæÒ»¶ÎÊÇ´òÂÒÑù±¾¼¯ [N,m1]=size(X1); [qs,qi]=sort(rand(N,1)); X1=X1(qi, ;y=y(qi, ;%Ê®±¶½»²æÑéÖ¤ indices = crossvalind('Kfold',N,10); for F = 1:10 test = (indices == F); train = ~test; Test1=X1(test, ;Train1=X1(train, ;ytrain=y(train);ytest=y(test); [n,m1]=size(Train1); D1=Train1*Train1'; e=ones(n,1); for c=1:10 C=2^c; O=zeros(n,1); cvx_begin variable lambda1(n); maximize(lambda1'*e-0.5*(lambda1.*ytrain)'*D1*(lambda1.*ytrain)); lambda1>=O; lambda1<=C*e; lambda1'*ytrain==0; cvx_end %ÏÂÃæÒ»¶ÎÊÇÇóÆ«ÖÃb£¬È¡À¸çÀÊÈÕ³Ë×Ó×î´óµÄ10¸öÖ§³ÖÏòÁ¿À´Ç󯽾ù¡£ [max,ind]=sort(lambda1); for j=1:10 bb(j)=(ytrain(ind(n-j+1))-(lambda1.*ytrain)'*Train1*Train1(ind(n-j+1), ');end b{c}=mean(bb); ypredict1{F,c}=Test1*Train1'*(lambda1.*ytrain)+b{c}; acc1(F,c)=accuracy(ytest,sign(ypredict1{F,c}+10e-10)); aveacc1=mean(acc1); end end |
» ²ÂÄãϲ»¶
Çóµ÷¼Á
ÒѾÓÐ4È˻ظ´
Ò»Ö¾Ô¸ Î÷±±´óѧ ×Ü·Ö282 Ó¢ÓïÒ»62 Çóµ÷¼Á
ÒѾÓÐ3È˻ظ´
µ÷¼Á310
ÒѾÓÐ4È˻ظ´
±±¿Æ281ѧ˶²ÄÁÏÇóµ÷¼Á
ÒѾÓÐ16È˻ظ´
Ò»Ö¾Ô¸ÉϺ£½»´óÉúÎïÓëҽҩר˶324·Ö£¬Çóµ÷¼Á
ÒѾÓÐ6È˻ظ´
Ò»Ö¾Ô¸¹þ¹¤´ó£¬085400£¬320£¬Çóµ÷¼Á
ÒѾÓÐ4È˻ظ´
316Çóµ÷¼Á
ÒѾÓÐ9È˻ظ´
085600 ²ÄÁÏÓ뻯¹¤ 329·ÖÇóµ÷¼Á
ÒѾÓÐ8È˻ظ´
0856Çóµ÷¼Á
ÒѾÓÐ6È˻ظ´
299Çóµ÷¼Á
ÒѾÓÐ4È˻ظ´

orcimbalance
ľ³æ (ÖøÃûдÊÖ)
- Ó¦Öú: 22 (СѧÉú)
- ½ð±Ò: 2057.1
- É¢½ð: 937
- ºì»¨: 3
- Ìû×Ó: 1306
- ÔÚÏß: 428.9Сʱ
- ³æºÅ: 1401449
- ×¢²á: 2011-09-14
- ÐÔ±ð: GG
- רҵ: ¼ÆËã»úÌåϵ½á¹¹
2Â¥2012-11-13 15:47:28













;
»Ø¸´´ËÂ¥
Äã°ÑSVMÔÀí¶¼ÅªÇå³þÁË£¿Ç¿£¡