Add files via upload

Backpropagation.m

@@ -0,0 +1,54 @@
+clc;
+clear all;
+v1=[0.6,-0.1,0.3];
+v2=[-0.3,0.4,0.5];
+w=[0.4,0.1,-0.2];
+t=[0,1,1,0];
+alpha=0.25;
+x=input('Enter input values: ');
+mse=1;
+while(mse>0.0005)
+    for i=1:4
+        zin1=x(i,:)*v1(1:2)'+v1(3);
+        zin2=x(i,:)*v2(1:2)'+v2(3);
+        z1=1./(1+exp(-zin1));
+        z2=1./(1+exp(-zin2));
+        yin=z1*w(1)+z2*w(2)+w(3);
+        y=1./(1+exp(-yin));
+
+        d0=(t(i)-y)*y*(1-y);
+        dt_w1=d0*z1*alpha;
+        dt_w2=d0*z2*alpha;
+        dt_w0=d0*alpha;
+        dt_w=[dt_w1 dt_w2 dt_w0];
+
+        din1=d0*w(1);
+        din2=d0*w(2);
+        d1=din1*z1*(1-z1);
+        d2=din2*z2*(1-z2);
+        dt_v11=d1*alpha*x(i,1);
+        dt_v21=d1*alpha*x(i,2);
+        dt_v01=d1*alpha;
+        dt_v12=d2*alpha*x(i,1);
+        dt_v22=d2*alpha*x(i,2);
+        dt_v02=d2*alpha;
+        dt_v1=[dt_v11 dt_v12 dt_v01];
+        dt_v2=[dt_v21 dt_v22 dt_v02];
+
+        w=w+dt_w;
+        v1=v1+dt_v1;
+        v2=v2+dt_v2;
+
+        mse=(t(i)-y).^2;
+    end
+end
+v1
+v2
+w
+for j = 1:4
+    zin=[x(j,:)*v1(1:2)'+v1(3),x(j,:)*v2(1:2)'+v2(3)];
+    z=1./(1+exp(-zin));
+    yin=z*w(1:2)'+w(3);
+    y(j)=1./(1+exp(-yin));
+end
+y

Defuzzification.m

@@ -0,0 +1,82 @@
+clc;
+clear all;
+
+x=0:0.01:10;
+y1=trapmf(x,[0 1 4 7]);
+y1=0.25*y1;
+y2=trapmf(x,[3 5 6 7]);
+y2=0.7*y2;
+y3=trapmf(x,[2 6 7 8]);
+y3=1*y3;
+figure(1)
+subplot(2,2,1);
+plot(x,y1,'linewidth',2);
+subplot(2,2,2);
+plot(x,y2,'linewidth',2);
+subplot(2,2,3);
+plot(x,y3,'linewidth',2);
+a=max(y1,y2);
+b=max(a,y3);
+subplot(2,2,4);
+plot(x,b,'linewidth',2);
+
+[max1,n]=max(b);
+first_max=(n-1)/100;
+for i=n:(size(x,2)+1)
+    if(b(i)~=max1)
+        last_max=(i-2)/100;
+        break;
+    end
+end
+first_max
+last_max
+mean_max=(first_max+last_max)/2
+
+sum1=sum(b);
+N=0;
+for i=0:size(x,2)-1
+    N=N+(b(i+1)*(i*0.01));
+end
+centroid=N/sum1
+
+[maxy1,n1]=max(y1);
+first_max1=(n1-1)/100;
+for i=n1:(size(x,2)+1)
+    if(y1(i)~=maxy1)
+        last_max1=(i-1)/100;
+        break;
+    end
+end
+mean_max1=(first_max1+last_max1)/2;
+
+[maxy2,n2]=max(y2);
+first_max2=(n2-1)/100;
+for i=n2:(size(x,2)+1)
+    if(y2(i)~=maxy2)
+        last_max2=(i-1)/100;
+        break;
+    end
+end
+mean_max2=(first_max2+last_max2)/2;
+
+[maxy3,n3]=max(y3);
+first_max3=(n3-1)/100;
+for i=n3:(size(x,2)+1)
+    if(y3(i)~=maxy3)
+        last_max3=(i-1)/100;
+        break;
+    end
+end
+mean_max3=(first_max3+last_max3)/2;
+num=mean_max1*max(y1)+mean_max2*max(y2)+mean_max3*max(y3);
+den=max(y1)+max(y2)+max(y3);
+wt_avg=num/den
+
+figure(2)
+plot(x,b,'linewidth',2);
+line([first_max first_max],[0 1],'color','red','linestyle','--','linewidth',2);
+line([last_max last_max],[0 1],'color','green','linestyle','--','linewidth',2);
+line([mean_max mean_max],[0 1],'color','yellow','linestyle','--','linewidth',2);
+line([centroid centroid],[0 1],'color','cyan','linestyle','--','linewidth',2);
+line([wt_avg wt_avg],[0 1],'color','magenta','linestyle','--','linewidth',2);
+legend('Membership Function Plot','First of Max','Last of Max','Mean of Max','Centroid','Weighted Average');

Kohonen Self Organizing Maps.m

@@ -0,0 +1,21 @@
+clc;
+clear all;
+pattern = input('Enter Training patterns: ');
+wt = input('Enter weights: ');
+alpha = input('Enter Learning Rate: ');
+epoch = 1000;
+for e=1:epoch
+    for i = 1:4
+        dist1 = (pattern(i,:)-wt(1,:)).^2;
+        dist1=sum(dist1);
+        dist2 = (pattern(i,:)-wt(2,:)).^2;
+        dist2=sum(dist2);
+        if(dist1<dist2)
+            wt(1,:)=wt(1,:)+alpha*(pattern(i,:)-wt(1,:));
+        else
+            wt(2,:)=wt(2,:)+alpha*(pattern(i,:)-wt(2,:));
+        end
+    end
+    alpha = alpha/2;
+end
+wt

LinearRegression.m

@@ -0,0 +1,32 @@
+clc;
+clear all;
+x = 0:0.01:0.99;
+t = 40*x+10+rand(1,100);
+w=0;
+b=0;
+epoch=60;
+r=0.25;
+for i=1:epoch
+    mse=0;
+    for j=1:100
+        y(j)=w*x(j)+b;
+        err=t(j)-y(j);
+        w=w+err*x(j)*r;
+        b=b+err*r;
+        mse=mse+err^2;
+    end
+    mse=mse/100;
+    mserror(i)=mse;
+end
+disp('Mean Square Error is:');
+disp(mse);
+figure(1)
+scatter(x,t);
+refline(w,b);
+xlabel('x');
+ylabel('t');
+title('Linear Regression');
+figure(2)
+plot(1:epoch,mserror);
+xlabel('No. of Iterations');
+ylabel('Mean Square Error');

Mamdani Fuzzy Model.m

@@ -0,0 +1,83 @@
+clc
+clear all
+
+figure(1)
+x=0:0.1:100;
+T1=trimf(x,[10 10 25]);
+plot(x,T1);
+hold on
+T2=trimf(x,[15 30 45]);
+plot(x,T2);
+hold on
+T3=trimf(x,[40 50 60]);
+plot(x,T3);
+hold on
+T4=trimf(x,[55 70 85]);
+plot(x,T4);
+hold on
+T5=trimf(x,[75 90 90]);
+plot(x,T5);
+
+figure(2)
+y=0:0.01:5;
+P1=trimf(y,[1 1 1.75]);
+plot(y,P1);
+hold on
+P2=trimf(y,[1.25 2 2.75]);
+plot(y,P2);
+hold on
+P3=trimf(y,[2 3 4]);
+plot(y,P3);
+hold on
+P4=trimf(y,[3.25 4 4.75]);
+plot(y,P4);
+hold on
+P5=trimf(y,[4.25 5 5]);
+plot(y,P5);
+
+figure(3)
+z=0:0.01:5;
+HP1=trimf(z,[1 1 1.5]);
+plot(z,HP1);
+hold on
+HP2=trimf(z,[1.25 2 2.75]);
+plot(z,HP2);
+hold on
+HP3=trimf(z,[2.5 3.125 3.75]);
+plot(z,HP3);
+hold on
+HP4=trimf(z,[3.5 4 4.5]);
+plot(z,HP4);
+hold on
+HP5=trimf(z,[4.25 5 5]);
+plot(z,HP5);
+
+T=T2(find(x==22.5,1,'first'));
+P=P2(find(y==1.5,1,'first'));
+HPMH=min(T,P);
+HP=HP4*HPMH;
+figure(4)
+plot(z,HP4)
+hold on
+area(z,HP)
+
+TL=T1(find(x==22.5,1,'first'));
+PL=P1(find(y==1.5,1,'first'));
+HPH=min(TL,PL);
+HPL=HP5*HPH;
+figure(5)
+plot(z,HP5)
+hold on
+area(z,HPL)
+
+figure(6)
+area(z,HP)
+hold on
+area(z,HPL)
+
+AreaMH=(1/2)*1*1
+AreaH=2*(1/2)*0.75*1
+CentreMH=4
+CentreH=5
+HPMH
+HPH

McCulloch Pitt Model.m

@@ -0,0 +1,50 @@
+clc;
+clear all;
+ip = input('Enter input values: ');
+ch = input('Enter choice of operation - 1. AND 2. NAND 3. OR. 4. NOR: ');
+wt = [1,1];
+switch ch
+    case 1
+        b = 2;
+        for i = 1:4
+            yin(i) = ip(i,:)*wt' - b;
+            if(yin(i)>=0)
+                yout(i) = 1;
+            else
+                yout(i) = 0;
+            end
+        end
+    case 2
+        b = 2;
+        for i = 1:4
+            yin(i) = ip(i,:)*wt' - b;
+            if(yin(i)>=0)
+                yout(i) = 0;
+            else
+                yout(i) = 1;
+            end
+        end
+    case 3
+        b = 1;
+        for i = 1:4
+            yin(i) = ip(i,:)*wt' - b;
+            if(yin(i)>=0)
+                yout(i) = 1;
+            else
+                yout(i) = 0;
+            end
+        end
+    case 4
+        b = 1;
+        for i = 1:4
+            yin(i) = ip(i,:)*wt' - b;
+            if(yin(i)>=0)
+                yout(i) = 0;
+            else
+                yout(i) = 1;
+            end
+        end
+end
+yout
+
+

MembershipFunctions.m

@@ -0,0 +1,14 @@
+clc;
+clear all;
+x=1:100;
+a=20;
+b=4;
+c=50;
+d=90;
+sigma=5;
+triangle=max(min((x-a)/(b-a),(c-x)/(c-b)),0);
+trap=max(min(min((x-a)/(b-a),1),(d-x)/(d-c)),0);
+gauss=exp(-((x-c)/sigma).^2*0.5);
+bell=1./((1+abs((x-c)/a)).^(2*b));
+figure(1)
+plot(x,bell);

Perceptron.m

@@ -0,0 +1,28 @@
+clc;
+clear all;
+w1=0;
+w2=0;
+b=0;
+wt=[w1 w2 b];
+ip=input('Enter inputs: ');
+alpha=input('Enter Learning Rate');
+flag=1;
+t=[1,0,0,0];
+while(flag==1)
+    for i=1:4
+        yin(i)=ip(i,:)*wt';
+        if(yin(i)>=0)
+            y(i)=1;
+        else
+            y(i)=0;
+        end
+        if(y(i)~=t(i))
+            err=t(i)-y(i);
+            wt=wt+err*alpha*ip(i,:);
+        end
+    end
+    if(y==t)
+        flag=0;
+    end
+end
+wt

Radial Basis Function Network.m

@@ -0,0 +1,21 @@
+clc;
+clear all;
+x=[0 0;0 1;1 0;1 1];
+t1=[1,1];
+t2=[0,0];
+for k=1:4
+    q=((x(k,1)-t1(1,1)).^2+(x(k,2)-t1(1,2)).^2);
+    G1(k)=exp(-q);
+end
+for k=1:4
+    q=((x(k,1)-t2(1,1)).^2+(x(k,2)-t2(1,2)).^2);
+    G2(k)=exp(-q);
+end
+G1
+G2
+G=[G1;G2;1 1 1 1]';
+D=[0 1 1 0]';
+W=inv(G'*G)*G'*D;
+a=G*W;
+a(a<0)=0;
+a