Thyroid class 1/2 results with SCG:

nhidden = 3;			% Number of hidden units.
alpha     = 0.00	;		% Coefficient of weight-decay prior. 
Nnoisepoints=12;  	% add some noise
VarNoise=0.01;	% with this variance
options(14) = 500	;		% Number of training cycles. 

    93     0					conf. 
     0   192

 N. err= 0,  Acc = 1

       92.994     0.031145		Soft Conf
     0.032898       191.99

        1107           9			noise confusion
          11        2293

 N. err= 20,  Acc = 0.99415  noise

==============================================================

Same with alpha     = 0.05

    93     0					conf. 
     0   192

 N. err= 0,  Acc = 1

      91.64       1.3552
    1.3692       190.62

   1105          11		noise conf
        6        2298

 N. err= 17,  Acc = 0.99503   noise
 
 ==============================================================
 
 Same with alpha     = 0.10
 
     92      1					conf. 
      0   192
 
  N. err= 1,  Acc = 0.99649
 
      90.919       2.0724
       2.0914        189.9
       
        1107           9
             5      2299  			noise conf
           
 N. err= 14,  Acc = 0.99591		noise
 
  ==============================================================
  
  Same with alpha     = 0.20
  
      92      1					conf. 
       0   192
  
   N. err= 1,  Acc = 0.99649
  
       90.111       2.9539
       3.0342       189.01
       
        1098          18
          11        2293				noise conf
            
   N. err= 29,  Acc = 0.99152		noise
   N. err= 19,  Acc = 0.99444  		spora wariancja
   
  ==============================================================
  
  Same with alpha     = 0.20, nhidd = 6
  
      92      1					conf. 
       1   191
  
N. err= 2,  Acc = 0.99298

     89.922       3.0652
       3.0942       188.89
       
        1098          18
          16        2288				noise conf
            
    N. err= 34,  Acc = 0.99006
    
  ==============================================================
  
  Same with alpha     = 0.20, nhidd = 1
  
      92      1					conf. 
       0   192
  
 N. err= 1,  Acc = 0.99649
 
       89.932       3.0681
       3.1309       188.87
       
         1103          13
          10        2294
          
N. err= 23,  Acc = 0.99327    
    
  ==============================================================
  
  Same with alpha     = 0.10, nhidd = 1
  
      92      1					conf. 
       0   192
  
 N. err= 1,  Acc = 0.99649

       90.933       2.0674
       2.0972        189.9
       
      1102          14
           8        2296
          
 N. err= 22,  Acc = 0.99357
 
  ==============================================================
   
   Same with alpha     = 0.00, nhidd = 1
   
       93      0					conf. 
        0   192
   
  N. err= 0,  Acc = 1

       92.989     0.010625
     0.010481       191.99
     
          1109           7
          10        2294
          
  N. err= 17,  Acc = 0.99503  

  ==============================================================
 
 30 h, a=0.05, noise 12 errors, good solution!