[Home]

**RPTSVM**

A Matlab code for recursive projection twin support vector machine.
(You could Right-Click [Code] , and Save, then you can download the whole matlab code.)

**Reference**

Yuan-Hai Shao, Zhen Wang, Wei-Jie Chen, Nai-Yang Deng*. A regularization for the projection twin support vector machine[J]. **Knowledge-Based Systems**, 2013, 37: 203¨C210.

**Main Function**

**function Predict_Y = RPTSVM(TestX,DataTrain,FunPara)**
**
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% RPTSVM: recursive projection twin support vector machine
%
% Predict_Y = RPTSVM(TestX,DataTrain,FunPara)
%
% Input:
% TestX - Test Data matrix. Each row vector of fea is a data point.
%
% DataTrain - Struct value in Matlab(Training data).
% DataTrain.A: Positive input of Data matrix.
% DataTrain.B: Negative input of Data matrix.
%
% FunPara - Struct value in Matlab. The fields in options that can be set:
% c1: [0,inf] Paramter to tune the weight.
% c2: [0,inf] Paramter to tune the weight.
% v1: [0,inf] Paramter to tune the weight.
% v2: [0,inf] Paramter to tune the weight.
% kerpara: Paramter of kernel function.
%
% Output:
% Predict_Y - Predict value of the TestX.
%
% Examples:
% DataTrain.A = rand(50,10);
% DataTrain.B = rand(60,10);
% TestX = rand(20,10);
% FunPara.c1=0.1;
% FunPara.c2=0.1;
% FunPara.v1=0.1;
% FunPara.v2=0.1;
% FunPara.kerpara.type='rbf'
% FunPara.kerpara.pars=1
% Predict_Y = RPTSVM(TestX,DataTrain,FunPara);
%
% Reference:
% Yuan-Hai Shao, Zhen Wang, Wei-Jie Chen, Nai-Yang Deng*.
% A regularization for the projection twin support vector machine[J].
% Knowledge-Based Systems,2013, 37: 203¨C210.
%
% Version 1.0 --Apr/2011
% Written by Yuan-Hai Shao (shaoyuanhai21@163.com)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
****
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Initailization
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
****
%tic;
inputA = DataTrain.A;
inputB = DataTrain.B;
m = size(TestX,1);
c1 = FunPara.c1;
c2 = FunPara.c2;
v1 = FunPara.v1;
v2 = FunPara.v2;
m1 = size(inputA,1);
m2 = size(inputB,1);
e1 = ones(m1,1);
e2 = ones(m2,1);
kerpara = FunPara.kerpara;
****
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Compute w1 and w2
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
****
if strcmp(kerpara.type,'rbf')
X = [DataTrain.A;DataTrain.B];
inputA = kernelfun(inputA,kerpara,X);
inputB = kernelfun(inputB,kerpara,X);
TestX = kernelfun(TestX,kerpara,X);
end
s1 = m1*cov(inputA,1);
s2 = m2*cov(inputB,1);
I = eye(length(s1));
E1 = v1*s1 + I;
E2 = v2*s2 + I;
H = inputB-1/m1*e2*e1'*inputA;
G = inputA-1/m2*e1*e2'*inputB;
****
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%RPTWSVM1
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
****
HE = E1\H';
HHE = H*HE;
HHE = (HHE+HHE')/2;
alpha = qpSOR(HHE,0.5,c1,0.05); %SOR
w1 = E1\H'*alpha;
****
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%RPTWSVM2
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
****
GE = E2\G';
GGE = G*GE;
GGE = (GGE+GGE')/2;
beta = qpSOR(GGE,0.5,c2,0.05); %SOR
w2 = E2\G'*beta;
****
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%Record w1 and w2
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
****
W1 = w1;
W2 = w2;
center1 = mean(inputA,1);
center2 = mean(inputB,1);
****
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%While multiple orthogonal recursive projection
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
****
% mop=0;
% while mop>0
% for i=1:m1
% inputA(i,:)=inputA(i,:)-(w1*inputA(i,:)*w1)';
% end
% for i=1:m2
% inputB(i,:)=inputB(i,:)-(w2*inputB(i,:)*w2)';
% end
% s1 = m1*cov(inputA,1);
% s2 = m2*cov(inputB,1);
% I = eye(length(s1));
% E1 = v1*s1 + I;
% E2 = v2*s2 + I;
% H = inputB-1/m1*e2*e1'*inputA;
% G = inputA-1/m2*e1*e2'*inputB;
% HE = E1\H';
% HHE = H*HE;
% HHE = (HHE+HHE')/2;
% alpha = qpSOR(HHE,0.5,c1,0.05); %SOR
% w1 = E1\H'*alpha;
% GE = E2\G';
% GGE = G*GE;
% GGE = (GGE+GGE')/2;
% beta = qpSOR(GGE,0.5,c2,0.05); %SOR
% w2 = E2\G'*beta;
% W1=[W1,w1];
% W2=[W2,w2];
% mop=mop-1;
% end
% toc
****
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Predict and output
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
****
Predict_Y = zeros(m1,1);
for i=1:m
Y11(i,:)=TestX(i,:)*W1-center1*W1;
Y22(i,:)=TestX(i,:)*W2-center2*W2;
if norm(Y11(i,:)) <= norm(Y22(i,:))
Predict_Y(i,:)=1;
else
Predict_Y(i,:)=-1;
end
end
**

Any question or advice please email to shaoyuanhai21@163.com.

- Last updated: Jun 5, 2013