matlab怎么實(shí)現(xiàn)基于廣義神經(jīng)網(wǎng)絡(luò)GRNN網(wǎng)絡(luò)入侵聚類

%% 清空環(huán)境文件

clear

clc

%% 提取攻擊數(shù)據(jù)

%攻擊樣本數(shù)據(jù)

load netattack;

P1=netattack;

T1=P1(:,39)';

P1(:,39)=[];

%數(shù)據(jù)大小

[R1,C1]=size(P1);

csum=20;  %提取訓(xùn)練數(shù)據(jù)多少

%% 模糊聚類

data=P1;

[center,U,obj_fcn] = fcm(data,5);    

for i=1:R1

    [value,idx]=max(U(:,i));

    a1(i)=idx;

end

%% 模糊聚類結(jié)果分析

Confusion_Matrix_FCM=zeros(6,6);

Confusion_Matrix_FCM(1,:)=0:5;

Confusion_Matrix_FCM(:,1)=[0:5]';

for nf=1:5

    for nc=1:5

        Confusion_Matrix_FCM(nf+1,nc+1)=length(find(a1(T1==nf)==nc));

    end

end

%% 網(wǎng)絡(luò)訓(xùn)練樣本提取

cent1=P1(a1==1,:);cent1=mean(cent1);

cent2=P1(a1==2,:);cent2=mean(cent2);

cent3=P1(a1==3,:);cent3=mean(cent3);

cent4=P1(a1==4,:);cent4=mean(cent4);

cent5=P1(a1==5,:);cent5=mean(cent5);

%提取范數(shù)最小為訓(xùn)練樣本

for n=1:R1

    ecent1(n)=norm(P1(n,:)-cent1);

    ecent2(n)=norm(P1(n,:)-cent2);

    ecent3(n)=norm(P1(n,:)-cent3);

    ecent4(n)=norm(P1(n,:)-cent4);

    ecent5(n)=norm(P1(n,:)-cent5);

end

for n=1:csum

    [~, me1]=min(ecent1);

    [~, me2]=min(ecent2);

    [~, me3]=min(ecent3);

    [~, me4]=min(ecent4);

    [va, me5]=min(ecent5);

    ecnt1(n,:)=P1(me1(1),:);ecent1(me1(1))=[];tcl(n)=1;

    ecnt2(n,:)=P1(me2(1),:);ecent2(me2(1))=[];tc2(n)=2;

    ecnt3(n,:)=P1(me3(1),:);ecent3(me3(1))=[];tc3(n)=3;

    ecnt4(n,:)=P1(me4(1),:);ecent4(me4(1))=[];tc4(n)=4;

    ecnt5(n,:)=P1(me5(1),:);ecent5(me5(1))=[];tc5(n)=5;

end

P2=[ecnt1;ecnt2;ecnt3;ecnt4;ecnt5];T2=[tcl,tc2,tc3,tc4,tc5];

k=0;

%% 迭代計(jì)算

for nit=1:10%開始迭代

    %% 廣義神經(jīng)網(wǎng)絡(luò)聚類

    net = newgrnn(P2',T2,50);   

    %訓(xùn)練廣義網(wǎng)絡(luò)

    a2=sim(net,P1') ;  

    %預(yù)測(cè)結(jié)果

    %輸出標(biāo)準(zhǔn)化（根據(jù)輸出來分類）

    a2(a2<=1.5)=1;

    a2(a2>1.5&a2<=2.5)=2;

    a2(a2>2.5&a2<=3.5)=3;

    a2(a2>3.5&a2<=4.5)=4;

    a2(a2>4.5)=5;

    %% 網(wǎng)絡(luò)訓(xùn)練數(shù)據(jù)再次提取

    cent1=P1(a2==1,:);cent1=mean(cent1);

    cent2=P1(a2==2,:);cent2=mean(cent2);

    cent3=P1(a2==3,:);cent3=mean(cent3);

    cent4=P1(a2==4,:);cent4=mean(cent4);

    cent5=P1(a2==5,:);cent5=mean(cent5);

    for n=1:R1%計(jì)算樣本到各個(gè)中心的距離

        ecent1(n)=norm(P1(n,:)-cent1);

        ecent2(n)=norm(P1(n,:)-cent2);

        ecent3(n)=norm(P1(n,:)-cent3);

        ecent4(n)=norm(P1(n,:)-cent4);

        ecent5(n)=norm(P1(n,:)-cent5);

    end

    %選擇離每類中心最近的csum個(gè)樣本

    for n=1:csum

        [~, me1]=min(ecent1);

        [~, me2]=min(ecent2);

        [~, me3]=min(ecent3);

        [~, me4]=min(ecent4);

        [va, me5]=min(ecent5);

        ecnt1(n,:)=P1(me1(1),:);ecent1(me1(1))=[];tc1(n)=1;

        ecnt2(n,:)=P1(me2(1),:);ecent2(me2(1))=[];tc2(n)=2;

        ecnt3(n,:)=P1(me3(1),:);ecent3(me3(1))=[];tc3(n)=3;

        ecnt4(n,:)=P1(me4(1),:);ecent4(me4(1))=[];tc4(n)=4;

        ecnt5(n,:)=P1(me5(1),:);ecent5(me5(1))=[];tc5(n)=5;

    end

    p2=[ecnt1;ecnt2;ecnt3;ecnt4;ecnt5];

    T2=[tc1,tc2,tc3,tc4,tc5];

    %統(tǒng)計(jì)分類結(jié)果

    Confusion_Matrix_GRNN=zeros(6,6);

    Confusion_Matrix_GRNN(1,:)=0:5;

    Confusion_Matrix_GRNN(:,1)=[0:5]';

    for nf=1:5

        for nc=1:5

            Confusion_Matrix_GRNN(nf+1,nc+1)=length(find(a2(T1==nf)==nc));

        end

    end

    pre2=0;

    for n=2:6

        pre2=pre2+max(Confusion_Matrix_GRNN(n,:));

    end

    pre2=pre2/R1*100;

end