% EM.m
%  EM algorithm for mixture of Gaussians
%  components
%    EM_init.m
%      initialization the probabilistic model parameters
%      Model = EM_init( m, x )
%      input
%         m : mixture number
%         x : reference data
%      output
%         Model : model structure
%
%    EM_dostep.m
%      do E_step and M_step and update model parameters
%      Model = EM_dostep( Model, x )
%      input
%         Model : model structure
%         x : data to learn
%      output
%         Model : model structure updated
%
%    EM_mixGaussPlot.m
%      2 dimensional plot for mixture of Gaussians
%      EM_mixGaussPlot( Model, x )
%      input
%         Model : Model to be illustrated
%         x : sample data to be illustrated

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Ndata　と　M　の値を変えて試して見よ
% 学習データ数
Ndata = 50;
% ユニットの個数
M = 3;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

load dataEM1

x=x(:,1:Ndata);

Model = EM_init( M, x );
EM_mixGaussPlot(Model,x)
hold off
for epoch = 1:50
  pause
  Model = EM_dostep( Model, x );
  EM_mixGaussPlot(Model,x)
  title( ['epoch=' num2str(epoch)])
  hold off
end