project2.asv

模式识别一份很好的作业,包括线性分类器；最小风险贝叶斯分类器；监督学习法分层聚类分析；K－L变换提取有效特征,支持向量机

load human;
load background;
for j=1:10
k_s=1;
k_d=120;
hm_tr(250,252)=0;
hm_test(250,252)=0;
[hm_tr,hm_test]=randQ(human);
randQ(background);
bg_tr(250,252)=0;
bg_test(250,252)=0;
[bg_tr,bg_test]=randQ(background);
M_hm=hm_tr(1,k_s:k_d);
for i=2:250
M_hm=M_hm+(hm_tr(i,k_s:k_d));
end
M_hm=M_hm/250;
M_bg=bg_tr(1,k_s:k_d);
for i=2:250
M_bg=M_bg+(bg_tr(i,k_s:k_d));
end
M_bg=M_bg/250;
E_hm(k_d-k_s+1,k_d-k_s+1)=0;
E_bg(k_d-k_s+1,k_d-k_s+1)=0;
for i=1:250
    E_hm=E_hm+(hm_tr(i,k_s:k_d)-M_hm)'*(hm_tr(i,k_s:k_d)-M_hm);
    E_bg=E_bg+(bg_tr(i,k_s:k_d)-M_bg)'*(bg_tr(i,k_s:k_d)-M_bg);
end
E_hm=E_hm/250;
E_bg=E_bg/250;
%test________________________________________________________
flag_hm(250)=0;
flag_bg(250)=0;
for i=1:250
    dk_hm=log(0.5)+log(0.95)-log(abs(det(E_bg)))/2-((hm_test(i,k_s:k_d)-M_bg)*inv(E_bg)*(hm_test(i,k_s:k_d)-M_bg)')/2;
    dk_bg=log(1.5)+log(0.05)-log(abs(det(E_hm)))/2-((hm_test(i,k_s:k_d)-M_hm)*inv(E_hm)*(hm_test(i,k_s:k_d)-M_hm)')/2;
    if(dk_hm<=dk_bg)
        flag_hm(i)=1;
    else flag_hm(i)=0;
    end
end
s2_hm=sum(flag_hm);
for i=1:250
    dk_hm=log(0.5)+log(0.95)-log(abs(det(E_bg)))/2-((bg_test(i,k_s:k_d)-M_bg)*inv(E_bg)*(bg_test(i,k_s:k_d)-M_bg)')/2;
    dk_bg=log(1.5)+log(0.05)-log(abs(det(E_hm)))/2-((bg_test(i,k_s:k_d)-M_hm)*inv(E_hm)*(bg_test(i,k_s:k_d)-M_hm)')/2;
    if(dk_hm>=dk_bg)
        flag_bg(i)=1;
    else flag_bg(i)=0;
    end
end
s2_bg=sum(flag_bg);
err_hg(j)=(250-s2_hm)/500;
err_gh(j)=(250-s2_bg)/500;
end
ERR_HG=0;
ERR_GH=0;
for j=1:10
    ERR_HG=ERR_HG+err_hg(j);
    ERR_GH=ERR_GH+err_gh(j);
end
ERR_HG=ERR_HG/10;
ERR_GH=ERR_GH/10;