neural_window.java

som 算法实现。在别的网站上下载的

package dm.clustering.KNN;

//this class holds the different types of neural nets
public class Neural_Window {

	public String neural_network_type;
	public int neural_network_number;
	
	public void display_menu_for_net_selection(int NNnum){
	  neural_network_number = NNnum;
	  System.out.println("****************");
	  System.out.println();
	  System.out.println( " Neural Network " + neural_network_number + " ");
	  System.out.println();
	  System.out.println();
	  System.out.println( "Please select one of the following network types from the Main Menu");
	  int i = 0;
	  do {System.out.println(); i = i + 1;} while (i < 3);
	  System.out.println("                            ");
	  System.out.println( " *** / Main Menu \\ ***");  
	  System.out.println();
	  System.out.println( " F.  Feedforward network using backpropagation " );
	  System.out.println( " A.  Adaptive Resonance Theory network for binary signals " );
	  System.out.println( " K.  Kohonen Self-Organizing Map " );
	  System.out.println( " R.  Radial Basis Function Network " );
	  System.out.println( " E.  Exit Program" );
	  System.out.println();
	  System.out.println( "Network Type (?) "); 
	  neural_network_type = MyInput.readString();
	  neural_network_type = neural_network_type.toUpperCase();
	  if(!neural_network_type.equalsIgnoreCase("E"))
	  {establish_network_type();}
	}
	
	private void establish_network_type(){
	   int NNN =  neural_network_number;

	   NeuralK KOH;

       //Kohonen Self-Organizing Map
	   KOH = new NeuralK();
	   Storage Kstore = new Storage();
	   KOH.construct_Kohonen_network();
	   KOH.network_training_testing(NNN);
	   Kstore.save_neural_network(KOH.Kohonen_Design);		   
    }
	
	public static void  main(String[] args){

	 int number_of_nets;
	 Neural_Window User_net = new Neural_Window();
	 System.out.println( " ******* Welcome to Pitt-Networks!! ******** " );
	 System.out.println( "Please enter the number of networks you wish to develop: "); 
	 number_of_nets = MyInput.readInt();

	 for(int NWnet = 1; NWnet < number_of_nets + 1; NWnet++)
	 {
	   User_net.display_menu_for_net_selection(NWnet);
	   if(User_net.neural_network_type.equalsIgnoreCase("E"))
	   {break;}
	 }	 
	}
}