您现在的位置:首页 >> ★免费资源 >> 源码下载 >> 内容

MATLAB代做-python代做-混合高斯em算法matlab源码可算三个参数

时间:2019-5-6 21:54:56 点击:

  核心提示:MATLAB代做-python代做-混合高斯em算法matlab源码可算三个参数...
function [W,M,V,L] = EM_GM_fast(X,k,ltol,maxiter,pflag,Init)
% [W,M,V,L] = EM_GM_fast(X,k,ltol,maxiter,pflag,Init) 
% EM algorithm for k multidimensional Gaussian mixture estimation
% (EM_GM_fast is the modified version of EM_GM for speed enchancement.
%  The functionalities of EM_GM_fast and EM_GM are identical.)
%
% Note: EM_GM_fast requires more memory than EM_GM to execute.
%       If EM_GM_fast does not provide any speed gain or is slower than EM_GM,
%       more memory is needed or EM_GM should be used instead.
%
% Inputs:
%   X(n,d) - input data, n=number of observations, d=dimension of variable
%   k - maximum number of Gaussian components allowed
%   ltol - percentage of the log likelihood difference between 2 iterations ([] for none)
%   maxiter - maximum number of iteration allowed ([] for none)
%   pflag - 1 for plotting GM for 1D or 2D cases only, 0 otherwise ([] for none)
%   Init - structure of initial W, M, V: Init.W, Init.M, Init.V ([] for none)
%
% Ouputs:
%   W(1,k) - estimated weights of GM
%   M(d,k) - estimated mean vectors of GM
%   V(d,d,k) - estimated covariance matrices of GM
%   L - log likelihood of estimates
%
% Written by
%   Patrick P. C. Tsui,
%   PAMI research group
%   Department of Electrical and Computer Engineering
%   University of Waterloo, 
%   March, 2006
%
%   Michael Boedigheimer
%   Amgen
%   Dept of Computational Biology
%   Thousand Oaks CA, 91320
%   Dec, 2005

%%%% Validate inputs %%%%
if nargin <= 1,
    disp('EM_GM must have at least 2 inputs: X,k!/n')
    return
elseif nargin == 2,
    ltol = 0.1; maxiter = 1000; pflag = 0; Init = [];
    err_X = Verify_X(X);
    err_k = Verify_k(k);
    if err_X | err_k, return; end
elseif nargin == 3,
    maxiter = 1000; pflag = 0; Init = [];
    err_X = Verify_X(X);
    err_k = Verify_k(k);
    [ltol,err_ltol] = Verify_ltol(ltol);    
    if err_X | err_k | err_ltol, return; end
elseif nargin == 4,
    pflag = 0;  Init = [];
    err_X = Verify_X(X);
    err_k = Verify_k(k);
    [ltol,err_ltol] = Verify_ltol(ltol);    
    [maxiter,err_maxiter] = Verify_maxiter(maxiter);
    if err_X | err_k | err_ltol | err_maxiter, return; end
elseif nargin == 5,
     Init = [];
    err_X = Verify_X(X);
    err_k = Verify_k(k);
    [ltol,err_ltol] = Verify_ltol(ltol);    
    [maxiter,err_maxiter] = Verify_maxiter(maxiter);
    [pflag,err_pflag] = Verify_pflag(pflag);
    if err_X | err_k | err_ltol | err_maxiter | err_pflag, return; end
elseif nargin == 6,
    err_X = Verify_X(X);
    err_k = Verify_k(k);
    [ltol,err_ltol] = Verify_ltol(ltol);    
    [maxiter,err_maxiter] = Verify_maxiter(maxiter);
    [pflag,err_pflag] = Verify_pflag(pflag);
    [Init,err_Init]=Verify_Init(Init);
    if err_X | err_k | err_ltol | err_maxiter | err_pflag | err_Init, return; end
else
    disp('EM_GM must have 2 to 6 inputs!');
    return
end

%%%% Initialize W, M, V,L %%%%
t = cputime;
if isempty(Init),  
    [W,M,V] = Init_EM(X,k); L = 0;    
else
    W = Init.W;
    M = Init.M;
    V = Init.V;
end
Ln = Likelihood(X,k,W,M,V); % Initialize log likelihood
Lo = 2*Ln;

%%%% EM algorithm %%%%
niter = 0;
while (abs(100*(Ln-Lo)/Lo)>ltol) & (niter<=maxiter),
    E = Expectation(X,k,W,M,V);     % E-step
    [W,M,V] = Maximization(X,k,E);  % M-step
    Lo = Ln;
    Ln = Likelihood(X,k,W,M,V);
    niter = niter + 1;
end 
L = Ln;

%%%% Plot 1D or 2D %%%%
if pflag==1,
    [n,d] = size(X);
    if d>2,
        disp('Can only plot 1 or 2 dimensional applications!/n');
    else
        Plot_GM(X,k,W,M,V);
    end
    elapsed_time = sprintf('CPU time used for EM_GM: %5.2fs',cputime-t);
    disp(elapsed_time); 
    disp(sprintf('Number of iterations: %d',niter-1));
end
%%%%%%%%%%%%%%%%%%%%%%
%%%% End of EM_GM %%%%
%%%%%%%%%%%%%%%%%%%%%%

function E = Expectation(X,k,W,M,V)
% This function is the modification of 'Expectation' in EM_GM made by 
% Mr. Michael Boedigheimer to enchance computational speed.
% Note: this modification requires more memory to execute.
%       If EM_GM_fast does not provide any speed gain or is slower than EM_GM,
%       more memory is needed or EM_GM should be used instead.
[n,d] = size(X);
E = zeros(n,k);
for j = 1:k,
    if V(:,:,j)==zeros(d,d), V(:,:,j)=ones(d,d)*eps; end
    E(:,j) = W(j).*mvnpdf( X, M(:,j)', V(:,:,j) );
end
total = repmat(sum(E,2),1,j);
E = E./total;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%% End of Expectation %%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%

function [W,M,V] = Maximization(X,k,E)
% This function is the modification of 'Maximization' in EM_GM made by 
% Mr. Michael Boedigheimer to enchance computational speed.
% Note: this modification requires more memory to execute.
%       If EM_GM_fast does not provide any speed gain or is slower than EM_GM,
%       more memory is needed or EM_GM should be used instead.
[n,d] = size(X);
W = sum(E);
M = X'*E./repmat(W,d,1);
for i=1:k,
    dXM = X - repmat(M(:,i)',n,1);
    Wsp = spdiags(E(:,i),0,n,n);
    V(:,:,i) = dXM'*Wsp*dXM/W(i);
end
W = W/n;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%% End of Maximization %%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

function L = Likelihood(X,k,W,M,V)
% Compute L based on K. V. Mardia, "Multivariate Analysis", Academic Press, 1979, PP. 96-97
% to enchance computational speed
[n,d] = size(X);
U = mean(X)';
S = cov(X);
L = 0;
for i=1:k,
    iV = inv(V(:,:,i));
    L = L + W(i)*(-0.5*n*log(det(2*pi*V(:,:,i))) ...
        -0.5*(n-1)*(trace(iV*S)+(U-M(:,i))'*iV*(U-M(:,i))));
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%% End of Likelihood %%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%

function err_X = Verify_X(X)
err_X = 1;
[n,d] = size(X);
if n<d,
    disp('Input data must be n x d!/n');
    return
end
err_X = 0;
%%%%%%%%%%%%%%%%%%%%%%%%%
%%%% End of Verify_X %%%%
%%%%%%%%%%%%%%%%%%%%%%%%%

function err_k = Verify_k(k)
err_k = 1;
if ~isnumeric(k) | ~isreal(k) | k<1,
    disp('k must be a real integer >= 1!/n');
    return
end
err_k = 0;
%%%%%%%%%%%%%%%%%%%%%%%%%
%%%% End of Verify_k %%%%
%%%%%%%%%%%%%%%%%%%%%%%%%

function [ltol,err_ltol] = Verify_ltol(ltol)
err_ltol = 1;
if isempty(ltol),
    ltol = 0.1;
elseif ~isreal(ltol) | ltol<=0,
    disp('ltol must be a positive real number!');
    return
end
err_ltol = 0;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%% End of Verify_ltol %%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%

function [maxiter,err_maxiter] = Verify_maxiter(maxiter)
err_maxiter = 1;
if isempty(maxiter),
    maxiter = 1000;
elseif ~isreal(maxiter) | maxiter<=0,
    disp('ltol must be a positive real number!');
    return
end
err_maxiter = 0;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%% End of Verify_maxiter %%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

function [pflag,err_pflag] = Verify_pflag(pflag)
err_pflag = 1;
if isempty(pflag),
    pflag = 0;
elseif pflag~=0 & pflag~=1,
    disp('Plot flag must be either 0 or 1!/n');
    return
end
err_pflag = 0;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%% End of Verify_pflag %%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

function [Init,err_Init] = Verify_Init(Init)
err_Init = 1;
if isempty(Init),
    % Do nothing;
elseif isstruct(Init),
    [Wd,Wk] = size(Init.W);
    [Md,Mk] = size(Init.M);
    [Vd1,Vd2,Vk] = size(Init.V);
    if Wk~=Mk | Wk~=Vk | Mk~=Vk,
        disp('k in Init.W(1,k), Init.M(d,k) and Init.V(d,d,k) must equal!/n')
        return
    end
    if Md~=Vd1 | Md~=Vd2 | Vd1~=Vd2,
        disp('d in Init.W(1,k), Init.M(d,k) and Init.V(d,d,k) must equal!/n')
        return
    end
else
    disp('Init must be a structure: W(1,k), M(d,k), V(d,d,k) or []!');
    return
end
err_Init = 0;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%% End of Verify_Init %%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%

function [W,M,V] = Init_EM(X,k)
[n,d] = size(X);
[Ci,C] = kmeans(X,k,'Start','cluster', ...
    'Maxiter',100, ...
    'EmptyAction','drop', ...
    'Display','off'); % Ci(nx1) - cluster indeices; C(k,d) - cluster centroid (i.e. mean)
while sum(isnan(C))>0,
    [Ci,C] = kmeans(X,k,'Start','cluster', ...
        'Maxiter',100, ...
        'EmptyAction','drop', ...
        'Display','off');
end
M = C';
Vp = repmat(struct('count',0,'X',zeros(n,d)),1,k);
for i=1:n, % Separate cluster points
    Vp(Ci(i)).count = Vp(Ci(i)).count + 1;
    Vp(Ci(i)).X(Vp(Ci(i)).count,:) = X(i,:);
end
V = zeros(d,d,k);
for i=1:k,
    W(i) = Vp(i).count/n;
    V(:,:,i) = cov(Vp(i).X(1:Vp(i).count,:));
end
%%%%%%%%%%%%%%%%%%%%%%%%
%%%% End of Init_EM %%%%
%%%%%%%%%%%%%%%%%%%%%%%%

function Plot_GM(X,k,W,M,V)
[n,d] = size(X);
if d>2,
    disp('Can only plot 1 or 2 dimensional applications!/n');
    return
end
S = zeros(d,k);
R1 = zeros(d,k);
R2 = zeros(d,k);
for i=1:k,  % Determine plot range as 4 x standard deviations
    S(:,i) = sqrt(diag(V(:,:,i)));
    R1(:,i) = M(:,i)-4*S(:,i);
    R2(:,i) = M(:,i)+4*S(:,i);
end
Rmin = min(min(R1));
Rmax = max(max(R2));
R = [Rmin:0.001*(Rmax-Rmin):Rmax];
clf, hold on
if d==1,
    Q = zeros(size(R));
    for i=1:k,
        P = W(i)*normpdf(R,M(:,i),sqrt(V(:,:,i)));
        Q = Q + P;
        plot(R,P,'r-'); grid on,
    end
    plot(R,Q,'k-');
    xlabel('X');
    ylabel('Probability density');
else % d==2
    plot(X(:,1),X(:,2),'r.');
    for i=1:k,
        Plot_Std_Ellipse(M(:,i),V(:,:,i));
    end
    xlabel('1^{st} dimension');
    ylabel('2^{nd} dimension');
    axis([Rmin Rmax Rmin Rmax])
end
title('Gaussian Mixture estimated by EM');
%%%%%%%%%%%%%%%%%%%%%%%%
%%%% End of Plot_GM %%%%
%%%%%%%%%%%%%%%%%%%%%%%%

function Plot_Std_Ellipse(M,V)
[Ev,D] = eig(V);
d = length(M);
if V(:,:)==zeros(d,d),
    V(:,:) = ones(d,d)*eps;
end
iV = inv(V);
% Find the larger projection
P = [1,0;0,0];  % X-axis projection operator
P1 = P * 2*sqrt(D(1,1)) * Ev(:,1);
P2 = P * 2*sqrt(D(2,2)) * Ev(:,2);
if abs(P1(1)) >= abs(P2(1)),
    Plen = P1(1);
else
    Plen = P2(1);
end
count = 1;
step = 0.001*Plen;
Contour1 = zeros(2001,2);
Contour2 = zeros(2001,2);
for x = -Plen:step:Plen,
    a = iV(2,2);
    b = x * (iV(1,2)+iV(2,1));
    c = (x^2) * iV(1,1) - 1;
    Root1 = (-b + sqrt(b^2 - 4*a*c))/(2*a);
    Root2 = (-b - sqrt(b^2 - 4*a*c))/(2*a);
    if isreal(Root1),
        Contour1(count,:) = [x,Root1] + M';
        Contour2(count,:) = [x,Root2] + M';
        count = count + 1;
    end
end
Contour1 = Contour1(1:count-1,:);
Contour2 = [Contour1(1,:);Contour2(1:count-1,:);Contour1(count-1,:)];
plot(M(1),M(2),'k+');
plot(Contour1(:,1),Contour1(:,2),'k-');
plot(Contour2(:,1),Contour2(:,2),'k-');
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%% End of Plot_Std_Ellipse %%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%


联系:highspeedlogic

QQ :1224848052

微信:HuangL1121

邮箱:1224848052@qq.com

网站:http://www.mat7lab.com/

网站:http://www.hslogic.com/

微信扫一扫:

作者:混合高斯em算法 来源:混合高斯em算法
  • 您是如何找到本站的?
  • 百度搜索
  • Google搜索
  • 查阅资料过程中
  • 论坛发现
  • 百度贴吧发现
  • 朋友介绍
本站最新成功开发工程项目案例
相关文章
  • 没有相关文章
相关评论
发表我的评论
  • 大名:
  • 内容:
  • matlab代做|matlab专业代做|matlab淘宝代做(www.hslogic.com) © 2019 版权所有 All Rights Reserved.
  • Email:highspeed_logic@163.com 站长QQ: 1224848052

    专业代做/代写/承接、MATLAB、SIMULINK、FPGA项目、博士/硕士/本科毕业设计、课题设计、论文,毕业论文,Coursework、Eassy、Assignment