translating a support vector machine structure into explicit form

Hi

I am new to SVM classification. I am trying to write a simple function to use the SVM struct in an explicit form. however using this function to classify new data doesn't yield the same result as SVM classify so I am assuming there is an error in my function. Can you help ? Thank you

The function is:

function f=classSVMU(x)

%shift [-0.136658807090128,-0.750778068467108]

%scaling factor [18.357296677281383,23.442173000970396]

x(1)=18.357296677281383*x(1) - 0.136658807090128*18.357296677281383;

x(2)=23.442173000970396*x(2) - 0.750778068467108*23.442173000970396;

%support vectors
SV=[1.54965309761314,0.155818518919654;-1.28999123215383,0.778501239257931;0.775204644040327,0.485474076745800;1.11940395673935,0.338960495489735;1.67651386968988,0.0884428172584560;-0.223206275049718,0.929632330370693;-0.274778194249021,0.980060247398309;0.614984820217217,0.629254293785940;-0.245359128459101,0.898846902484522;-1.69540650907984,0.600598657976632;-1.03118526557023,0.753173782386330;1.02599894259038,0.319833849471352;1.04876917164531,0.281621620945848;0.944516027577421,0.358483808260766;-0.0778111902177782,0.830042432487916];

alpha=[-1.06485892316272;-0.226204522991954;0.650000000000000;-0.655214684609375;0.650000000000000;0.650000000000000;-1.54628923267320;-1.99064892883321;0.650000000000000;0.0471015326634750;0.287562832989304;0.650000000000000;0.650000000000000;0.650000000000000;0.598551926617678];

%bias
f=5.060676156772543;

for i=1:15 %15 SV
    %poly kernel of order 5
    f=f+alpha(i)*(SV(i,:)*x'+1)^5;

end

Answer 1

[Ilya]

See this thread.

[Davout]

Thank you. It did work. I wanted to make a mex file out of it which does not seem to be possible with svmclassify.