svrplot.m

支持向量机的一个训练软件

function [x,z]=svrplot(X,Y,ker,beta,bias,epsi,par);
%This is a routine to plot 1D regression problem with SVM
%
%  [x,z]=svrplot(X,Y,ker,beta,bias,e);
%
%  Parameters: 
%
%              X: matrix of input vectors. Each row of x1 represents a d-dimensional vector. 
%                The number of rows is the number of training samples.
%              Y: column vector of the targets values. Each row represent the taerget of each sample in X.
%              ker: the kernel to be employed. It has to be a string out of: 'linear', 'poly_h', 'poly_i' and 'rbf'.
%                Type help kernel for further information.
%              beta: the value of the alphas-alpha*. (the second output of irwls1.m or irwls3.m)
%              bias: Bias term.  (the third output of irwls1.m or irwls3.m).
%              epsi: is the epsilon insensitive zone defined arround the regression estimate. 
%               (The 5th input to irwls1.m or the 4th output of irwls3.m)
%              par: is the parameter of the used kernel. type help kernel for further information.
%                   (The same value used for the 6th input parater in irwls.m).
%
%  Author: Steve Gunn ([e-mail])
%  Modified by: Fernando Perez-Cruz ([e-mail])

  if (nargin < 5 | nargin > 7) % check correct number of arguments
    help svrplot
  else

    epsilon = 1e-4;  
    
   if((nargin==5 | nargin==6) & ker(1)~='l')
       disp('The used kernel needs an input parameter.');
       return
   elseif(nargin==6 | nargin==5)
       par=0;
   end
   
   if(nargin==5)
       epsi=0;       
   end
   
    
    xmin = min(X(:,1));, xmax = max(X(:,1)); 
    xa = (xmax - xmin);

    % Plot function value

    n = size(X,1);
    x = [xmin:(xmax-xmin)/300:xmax]; 
    z = bias*ones(size(x));
    
    Ind=find(abs(beta)>epsilon);
    for x1 = 1 : length(x)
      z(x1)=kernel(ker,x(x1),X(Ind,:),par)*beta(Ind)+bias;
    end
    plot(x,z,'k','linewidth',2);
    axis off
    hold on
    
    % Plot epsi insensitive zone
    if (epsi > 0)
       plot(x,z+epsi,'b:')
       plot(x,z-epsi,'b:')
    end
 
    % Plot Training points
    
    for i = 1:size(Y)
      plot(X(i),Y(i),'k+') 
      if (abs(beta(i)) > epsilon)
        plot(X(i),Y(i),'ro','markersize',10) % Support Vector
      end
    end
    set(gca,'XLim',[xmin xmax]);  
    hold off

  end