Merge pull request #35 from sofia100/main

soft computing algos

hebbLearning.cpp

@@ -0,0 +1,76 @@
+#include<iostream>
+using namespace std;
+
+int main()
+{
+	int  m,n;
+	cout<<"enter no.of features and no.of training datasets: \n";
+	cin>> m>>n;
+
+	int wt1[m], wt2[m];
+
+	int input[n][m];
+	cout<<"enter the input matrix row wise "<<endl;
+
+	for(int i=0;i<n;i++)
+	{
+		for(int j=0;j<m;j++)
+		{
+			cin>>input[i][j];
+		}
+	}
+
+	int target1[n], target2[n];
+
+	cout<<" Enter the target in binary : "<<endl;
+	for(int i=0;i<n;i++)
+	{
+		cin>>target1[i];
+	}
+
+	cout<<"Enter the target in bipolar: "<<endl;
+	for(int i=0;i<n;i++)
+	{
+		cin>>target2[i];
+	}
+
+	for(int i=0;i<m;i++)
+	{
+		wt1[i]=0; //step 1: initialise all wts to 0
+		wt2[i]=0;
+	}
+
+	for(int j=0;j<n;j++) //step 2: for each training vector on target pair do steps 3 to 5
+	{//step 3: x_0 = 1 and xi = si
+		cout<<"###########j="<<j<<endl;
+		for(int i=0; i<m;i++)
+		{
+			//step 4: y_out = T_i 
+			//step 5: adjust the weights as WT_new= WT_old + (x_i*y_out)
+			wt1[i]+= (input[j][i]*target1[j]);
+			cout<<"weight1 at i="<<i<<" is "<<wt1[i]<<endl;
+			wt2[i]+= (input[j][i]*target2[j]);
+			cout<<"wt2 at i="<<i<<" is "<<wt2[i]<<endl;
+
+		}
+
+	}
+	cout<<"************OUTPUT**************\nafter 1 epoch: binary weights: "<<endl;
+
+	for (int i = 0; i < m; ++i)
+	{
+		/* code */
+		cout<<wt1[i]<<" ";
+	}
+	cout<<"\nafter 1 epoch: bipolar weights: "<<endl;
+
+	for (int i = 0; i < m; ++i)
+	{
+		/* code */
+		cout<<wt2[i]<<" ";
+	}
+
+
+
+	return 0;
+}

perceptron.cpp

@@ -0,0 +1,99 @@
+//given by Rosenblat
+//inspiration from retina of eye
+//seems like single feed forward network
+#include<iostream>
+using namespace std;
+
+int main()
+{
+	int  m,n;
+	float lr, t;
+	cout<<"enter no.of features , no.of training datasets , threshold and learning rate : \n";
+	cin>> m>>n>>t>>lr;
+
+	int wt2[m];
+
+	int input[n][m];
+	cout<<"enter the input matrix row wise "<<endl;
+
+	for(int i=0;i<n;i++)
+	{
+		for(int j=0;j<m;j++)
+		{
+			cin>>input[i][j];
+		}
+	}
+
+	int  target2[n];
+
+
+
+	cout<<"Enter the target in bipolar: "<<endl;
+	for(int i=0;i<n;i++)
+	{
+		cin>>target2[i];
+	}
+
+	for(int i=0;i<m;i++)
+	{
+		//step 1: initialise all wts to 0
+		wt2[i]=0;
+	}
+	int y_out;
+
+	for(int j=0;j<n;j++)
+	{
+		cout<<"\n########### training set j="<<j<<endl;
+int y_in=0;
+		for(int i=0; i<m;i++)
+		{
+			 y_in+= wt2[i]*input[j][i];
+			}
+			 cout<< "y_in = "<<y_in<<endl;
+
+			 if(y_in > t)
+			 	y_out=1;
+			 else if(y_in < -1*t)
+			 	y_out=-1;
+			 else
+			 	y_out=0;
+//			 cout<<"wt2 at i="<<i<<" is "<<wt2[i]<<endl;
+
+			 cout<< "y_out = "<<y_out<<endl;
+
+
+			 if(y_out != target2[j])
+			 {
+			 			for(int i=0; i<m;i++)
+
+			 				wt2[i]+= (lr *target2[j] *input[j][i]);
+			 			cout<<"updated weights :\n";
+			 for (int i = 0; i < m; ++i)
+	{
+		/* code */
+		cout<<wt2[i]<<" ";
+	}
+
+			 }
+
+			 else
+			 	{cout<<"\ntarget and y_out are equal. stop the training";
+			 	break;
+			 }
+
+
+
+	}
+	cout<<"\n************OUTPUT**************\nafter 1 epoch:  weights: "<<endl;
+
+
+	for (int i = 0; i < m; ++i)
+	{
+		/* code */
+		cout<<wt2[i]<<" ";
+	}
+
+
+
+	return 0;
+}