The Spider Objects

 
 
 
 
   
  Pre-Image Technique using MDS by Kwok et al.
   
 


   a = pmg_mds(alg,hyper)
   generates a pre-image object; uses the MDS Method
   -------------------------------------------------------
   hyperparameters:
   child=kernel     kernel to work with.
   map='rbf_inv'	   function to convert kernel distance back to input distance
   eps=1e-5         epsilon
   n=5         	   number of nearest neigbhors
   kp=kpca		   kpca object
   dn=sample data   data pool, if empty uses kp.dat		
   stats:
   methods:
   train         calls test
   test			find the preimage 
  
   example:
   k=kernel('rbf',1.5);
   
   d=gen(toy2d);
   d2=gen(toy2d);
   [r,kp]=train(kpca({k,'feat=20'}),d2);
   p0=pmg_mds;
   p0.kp=kp; 
   p0.child=k;
   p0.n=10;
   
   
   r=test(kp,d);
   
   reconstruct=test(p0,r)
   
   clf;
   hold on;
   plot(d2.X(:,1),d2.X(:,2),'g.');
   plot(d.X(:,1),d.X(:,2),'r.'); 
   plot(reconstruct.X(:,1),reconstruct.X(:,2),'o');

Reference : The Pre-Image Problem in Kernel Methods

Author : James Kwok et al.