gmm_classify.m

一个matlab的工具包,里面包括一些分类器 例如 KNN KMEAN SVM NETLAB 等等有很多.

function  [Y_compute, Y_prob] = GMM_classify(para, X_train, Y_train, X_test, Y_test, num_class)

Y_compute = zeros(size(Y_test)); Y_prob = zeros(size(Y_test));
if (isempty(X_train)),
   fprintf('Error: The training set is empty!\n');
   return;
end;

[class_set, num_class] = GetClassSet(Y_train);
if (nargin <= 5)
    num_class = 2;
end;

p = str2num(char(ParseParameter(para, {'-NumMix'}, {'1'})));
num_mix = p(1);
num_feature = size(X_train, 2);

% Fix seeds for reproducible results
randn('state', 42);
rand('state', 42);

for i = 1:num_class    
    % Convert the binary labels into +/-1
    data =  X_train(Y_train == class_set(i),:);
    
    mix = gmm(num_feature, num_mix, 'diag');
    
    options = foptions;
    options(14) = 5;	% Just use 5 iterations of k-means in initialisation
    % Initialise the model parameters from the data
    mix = gmminit(mix, data, options);

    % Set up vector of options for EM trainer
    options = zeros(1, 18);
    options(1)  = 0;		% Prints out error values.
    options(14) = 20;		% Max. Number of iterations.
    [mix, options, errlog] = gmmem(mix, data, options);
    
    Y_prob_matrix(:, i) = gmmprob(mix, X_test);
end;

[Y_prob Index] = max(Y_prob_matrix, [], 2);
Y_compute = class_set(Index);