bayescls.m

patten regnization source从1-14章能运行

function [y, dfce] = bayescls( X, model )
% BAYESCLS 啊快矫救 葛胆俊 措茄 海捞矫救 盒幅扁
% 
% 汲疙:
%  捞 窃荐绰 啊快矫救 葛胆俊 蝶扼 涝仿 氦磐 X俊 措茄 荤饶犬伏阑 拌魂窍绊 
%  涝仿 氦磐甸篮 荤饶 犬伏蔼捞 啊厘 奴 努贰胶肺 盒幅 登绢柳促. 
%  
% 涝仿:
%  X [dim x num_data] 盒幅窍绊历窍绰 涝仿 氦磐 
%
%  model.Mu [dim x num_classes] 啊快矫救 葛胆狼 乞闭 氦磐.
%  model.C [dim x dim x num_classes] 啊快矫救 葛胆狼 傍盒魂 青纺.
%  model.P [1 x num_classes] 荤傈 犬伏蔼.
%
% 免仿:
%  y [1 x num_data] 扼骇 
%  dfce [num_classes x num_data] 犬伏蔼 P(x|y)*P(y), 
%   咯扁辑,  x绰 盒幅等 氦磐捞绊 y绰 努贰胶 侥喊磊捞促. 
%

[dim,num_data]=size(X);
num_classes = size(model.Mu,2);

dfce=zeros(num_classes,num_data);

for i=1:num_classes,
  nconst = 1/((2*pi)^(dim/2) * sqrt(det(model.C(:,:,i))));
    
  dfce(i,:) = ...
     model.P(i)*nconst*exp(-0.5*mahalan(X,model.Mu(:,i),model.C(:,:,i)));
end

[tmp,y] = max(dfce);

return;