fix email

chapter01/entropy.m

@@ -1,6 +1,6 @@
 function z = entropy(x)
 % Compute entropy H(x) of a discrete variable x.
-% Written by Mo Chen (mochen80@gmail.com).
+% Written by Mo Chen (sth4nth@gmail.com).
 n = numel(x);
 x = reshape(x,1,n);
 [u,~,label] = unique(x);

chapter01/jointEntropy.m

@@ -1,6 +1,6 @@
 function z = jointEntropy(x, y)
 % Compute joint entropy H(x,y) of two discrete variables x and y.
-% Written by Mo Chen (mochen80@gmail.com).    
+% Written by Mo Chen (sth4nth@gmail.com).    
 assert(numel(x) == numel(y));
 n = numel(x);
 x = reshape(x,1,n);

chapter01/mutInfo.m

@@ -1,6 +1,6 @@
 function z = mutInfo(x, y)
 % Compute mutual information I(x,y) of two discrete variables x and y.
-% Written by Mo Chen (mochen80@gmail.com).
+% Written by Mo Chen (sth4nth@gmail.com).
 assert(numel(x) == numel(y));
 n = numel(x);
 x = reshape(x,1,n);

chapter01/relatEntropy.m

@@ -1,6 +1,6 @@
 function z = relatEntropy (x, y)
 % Compute relative entropy (a.k.a KL divergence) KL(p(x)||p(y)) of two discrete variables x and y.
-% Written by Mo Chen (mochen80@gmail.com).    
+% Written by Mo Chen (sth4nth@gmail.com).    
 assert(numel(x) == numel(y));
 n = numel(x);
 x = reshape(x,1,n);

chapter02/pdfDirichletLn.m

@@ -3,7 +3,7 @@
 %   X: d x n data matrix satifying (sum(X,1)==ones(1,n) && X>=0)
 %   a: d x k parameters
 %   y: k x n probability density
-% Written by Mo Chen (mochen80@gmail.com).
+% Written by Mo Chen (sth4nth@gmail.com).
 X = bsxfun(@times,X,1./sum(X,1));
 if size(a,1) == 1
     a = repmat(a,size(X,1),1);

chapter02/pdfGaussLn.m

@@ -1,6 +1,6 @@
 function y = pdfGaussLn(X, mu, sigma)
 % Compute log pdf of a Gaussian distribution.
-% Written by Mo Chen (mochen80@gmail.com).
+% Written by Mo Chen (sth4nth@gmail.com).
 
 [d,n] = size(X);
 k = size(mu,2);

chapter02/pdfKdeLn.m

@@ -1,6 +1,5 @@
 function z = pdfKdeLn (X, Y, sigma)
 % Compute log pdf of kernel density estimator.
-% Written by Mo Chen ([email protected]).
-    D = bsxfun(@plus,full(dot(X,X,1)),full(dot(Y,Y,1))')-full(2*(Y'*X));
-    z = logSumExp(D/(-2*sigma^2),1)-0.5*log(2*pi)-log(sigma*size(Y,2));
-endfunction
+% Written by Mo Chen ([email protected]).
+D = bsxfun(@plus,full(dot(X,X,1)),full(dot(Y,Y,1))')-full(2*(Y'*X));
+z = logSumExp(D/(-2*sigma^2),1)-0.5*log(2*pi)-log(sigma*size(Y,2));

chapter02/pdfMnLn.m

@@ -1,6 +1,6 @@
 function z = pdfMnLn (x, p)
 % Compute log pdf of a multinomial distribution.
-% Written by Mo Chen (mochen80@gmail.com).    
+% Written by Mo Chen (sth4nth@gmail.com).    
     if numel(x) ~= numel(p)
         n = numel(x);
         x = reshape(x,1,n);

chapter02/pdfStLn.m

@@ -1,6 +1,6 @@
 function y = pdfStLn(X, mu, sigma, v)
 % Compute log pdf of a student-t distribution.
-% Written by mo Chen (mochen80@gmail.com).
+% Written by mo Chen (sth4nth@gmail.com).
 [d,k] = size(mu);
 
 if size(sigma,1)==d && size(sigma,2)==d && k==1

chapter02/pdfWishartLn.m

@@ -1,6 +1,6 @@
 function y = pdfWishartLn(Sigma, v, W)
 % Compute log pdf of a Wishart distribution.
-% Written by Mo Chen (mochen80@gmail.com).
+% Written by Mo Chen (sth4nth@gmail.com).
 d = length(Sigma);
 B = -0.5*v*logdet(W)-0.5*v*d*log(2)-logmvgamma(0.5*v,d);
 y = B+0.5*(v-d-1)*logdet(Sigma)-0.5*trace(W\Sigma);

chapter02/pdflogVmfLn.m

@@ -1,6 +1,6 @@
 function y = pdflogVmfLn(X, mu, kappa)
 % Compute log pdf of a von Mises-Fisher distribution.
-% Written by Mo Chen (mochen80@gmail.com).
+% Written by Mo Chen (sth4nth@gmail.com).
 d = size(X,1);
 c = (d/2-1)*log(kappa)-(d/2)*log(2*pi)-logbesseli(d/2-1,kappa);
 q = bsxfun(@times,mu,kappa)'*X;

chapter03/linInfer.m

@@ -2,6 +2,7 @@
 % Compute linear model reponse y = w'*x+b and likelihood
 % X: d x n data
 % t: 1 x n response
+% Written by Mo Chen ([email protected]).
 w = model.w;
 b = model.w0;
 y = w'*X+b;

chapter03/regress.m

@@ -2,6 +2,7 @@
 % Fit linear regression model t=w'x+b
 % X: d x n data
 % t: 1 x n response
+% Written by Mo Chen ([email protected]).
 if nargin < 3
     lambda = 0;
 end

chapter03/regressEbEm.m

@@ -2,6 +2,7 @@
 % Fit empirical Bayesian linear model with EM
 % X: d x n data
 % t: 1 x n response
+% Written by Mo Chen ([email protected]).
 if nargin < 3
     alpha = 0.02;
     beta = 0.5;

chapter03/regressEbFp.m

@@ -2,6 +2,7 @@
 % Fit empirical Bayesian linear model with Mackay fixed point method
 % X: d x n data
 % t: 1 x n response
+% Written by Mo Chen ([email protected]).
 if nargin < 3
     alpha = 0.02;
     beta = 0.5;

chapter04/classFda.m

@@ -1,5 +1,6 @@
 function U = classFda(X, y, d)
 % Fisher (linear) discriminant analysis
+% Written by Mo Chen ([email protected]).
 n = size(X,2);
 k = max(y);
 

chapter04/classLogitBin.m

@@ -1,5 +1,6 @@
 function [model, llh] = classLogitBin(X, t, lambda)
 % logistic regression for binary classification (Bernoulli likelihood)
+% Written by Mo Chen ([email protected]).
 if any(unique(t) ~= [0,1])
     error('t must be a 0/1 vector!');
 end

chapter04/classLogitMul.m

@@ -1,5 +1,6 @@
 function [model, llh] = classLogitMul(X, t, lambda, method)
 % logistic regression for multiclass problem (Multinomial likelihood)
+% Written by Mo Chen ([email protected]).
 if nargin < 4
     method = 1;
 end

chapter04/sigmoid.m

@@ -1,2 +1,3 @@
 function y = sigmoid(x)
+% Written by Mo Chen ([email protected]).
 y = 1./(1+exp(-x));

chapter04/softmax.m

@@ -1,7 +1,7 @@
 function s = softmax(x, dim)
 % Compute softmax
 %   By default dim = 1 (columns).
-% Written by Michael Chen ([email protected]).
+% Written by Mo Chen ([email protected]).
 if nargin == 1, 
     % Determine which dimension sum will use
     dim = find(size(x)~=1,1);

chapter06/knCenterize.m

@@ -1,5 +1,6 @@
 function Kc = knCenterize(kn, X, Xt)
 % Centerize the data in the kernel space
+% Written by Mo Chen ([email protected]).
 K = kn(X,X);
 mK = mean(K);
 mmK = mean(mK);

chapter06/knGauss.m

@@ -1,5 +1,6 @@
 function K = knGauss(X, Y, s)
 % Gaussian (RBF) kernel
+% Written by Mo Chen ([email protected]).
 if nargin < 3
     s = 1;
 end

chapter06/knInfer.m

@@ -1,5 +1,6 @@
 function [y, sigma, p] = knInfer(x, model)
 % inference for kernel model
+% Written by Mo Chen ([email protected]).
 kn = model.kn;
 a = model.a;
 X = model.X;

chapter06/knLin.m

@@ -1,4 +1,5 @@
 function K = knLin(X, Y)
 % Linear kernel (inner product)
+% Written by Mo Chen ([email protected]).
 K = X'*Y;
 

chapter06/regressKn.m

@@ -1,5 +1,6 @@
 function model = regressKn(X, t, lambda, kn)
 % Gaussian process for regression
+% Written by Mo Chen ([email protected]).
 if nargin < 4
     kn = @knGauss;
 end

chapter07/classRvmEbEm.m

@@ -1,6 +1,7 @@
 function [model, llh] = classRvmEbEm(X, t, alpha)
 % Relevance Vector Machine classification training by empirical bayesian (ARD)
 % using standard EM update 
+% Written by Mo Chen ([email protected]).
 if nargin < 3
     alpha = 0.02;
 end

chapter07/classRvmEbFp.m

@@ -1,6 +1,7 @@
 function [model, llh] = classRvmEbFp(X, t, alpha)
 % Relevance Vector Machine classification training by empirical bayesian (ARD)
 % using fix point update (Mackay update)
+% Written by Mo Chen ([email protected]).
 if nargin < 3
     alpha = 0.02;
 end

chapter07/optLogitNewton.m

@@ -4,7 +4,7 @@
 % t: 1 x n 0/1 label
 % A: d x d regularization penalty
 % w: d x 1 initial value of w
-
+% Written by Mo Chen ([email protected]).
 [d,n] = size(X);
 
 if nargin < 4

chapter07/regressRvmEbCd.m

@@ -3,6 +3,7 @@
 % Analysis of sparse Bayesian learning. NIPS(2002). By Faul and Tipping
 % Fast marginal likelihood maximisation for sparse Bayesian models.
 % AISTATS(2003). by Tipping and Faul 
+% Written by Mo Chen ([email protected]).
 [d,n] = size(X);
 xbar = mean(X,2);
 tbar = mean(t,2);

chapter07/regressRvmEbEm.m

@@ -1,6 +1,7 @@
 function [model, llh] = regressRvmEbEm(X, t, alpha, beta)
 % Relevance Vector Machine regression training by empirical bayesian (ARD)
 % using standard EM update 
+% Written by Mo Chen ([email protected]).
 if nargin < 3
     alpha = 0.02;
     beta = 0.5;

chapter07/regressRvmEbFp.m

@@ -1,6 +1,7 @@
 function [model, llh] = regressRvmEbFp(X, t, alpha, beta)
 % Relevance Vector Machine regression training by empirical bayesian (ARD)
 % using fix point update (Mackay update)
+% Written by Mo Chen ([email protected]).
 if nargin < 3
     alpha = 0.02;
     beta = 0.5;

chapter07/regressRvmEbFpSvd.m

@@ -1,6 +1,7 @@
 function [model, llh] = regressRvmEbFpSvd(X, t, alpha, beta)
 % Relevance Vector Machine regression training by empirical bayesian (ARD)
 % using fix point update (Mackay update) with SVD
+% Written by Mo Chen ([email protected]).
 if nargin < 3
     alpha = 0.02;
     beta = 0.5;

chapter09/clusterKmeans.m

@@ -2,7 +2,7 @@
 % Perform k-means clustering.
 %   X: d x n data matrix
 %   k: number of seeds
-% Written by Michael Chen ([email protected]).
+% Written by Mo Chen ([email protected]).
 n = size(X,2);
 last = 0;
 label = ceil(k*rand(1,n));  % random initialization

chapter09/mixBernoulliEm.m

@@ -2,7 +2,7 @@
 % Perform EM algorithm for fitting the Bernoulli mixture model.
 %   X: d x n data matrix
 %   init: k (1 x 1) or label (1 x n, 1<=label(i)<=k) or center (d x k)
-% Written by Michael Chen ([email protected]).
+% Written by Mo Chen ([email protected]).
 %% initialization
 fprintf('EM for mixture model: running ... \n');
 n = size(X,2);

chapter09/mixGaussEm.m

@@ -2,7 +2,7 @@
 % Perform EM algorithm for fitting the Gaussian mixture model.
 %   X: d x n data matrix
 %   init: k (1 x 1) or label (1 x n, 1<=label(i)<=k) or center (d x k)
-% Written by Michael Chen ([email protected]).
+% Written by Mo Chen ([email protected]).
 %% initialization
 fprintf('EM for Gaussian mixture: running ... \n');
 R = initialization(X,init);

chapter09/mixMnEm.m

@@ -2,7 +2,7 @@
 % Perform EM algorithm for fitting the multinomial mixture model.
 %   X: d x n data matrix
 %   init: k (1 x 1) or label (1 x n, 1<=label(i)<=k) or center (d x k)
-% Written by Michael Chen ([email protected]).
+% Written by Mo Chen ([email protected]).
 %% initialization
 fprintf('EM for mixture model: running ... \n');
 n = size(X,2);

chapter10/mixGaussVb.m

@@ -3,8 +3,7 @@
 %   X: d x n data matrix
 %   init: k (1 x 1) or label (1 x n, 1<=label(i)<=k) or center (d x k)
 % Reference: Pattern Recognition and Machine Learning by Christopher M. Bishop (P.474)
-% Written by Michael Chen ([email protected]).
-
+% Written by Mo Chen ([email protected]).
 fprintf('Variational Bayesian Gaussian mixture: running ... \n');
 [d,n] = size(X);
 if nargin < 3

chapter10/regressRvmVb.m

@@ -2,6 +2,7 @@
 % Fit empirical Bayesian linear model with EM
 % X: m x n data
 % t: 1 x n response
+% Written by Mo Chen ([email protected]).
 [m,n] = size(X);
 if nargin < 3
     a0 = 1e-4;

chapter10/regressVb.m

@@ -2,6 +2,7 @@
 % Fit empirical Bayesian linear model with EM
 % X: m x n data
 % t: 1 x n response
+% Written by Mo Chen ([email protected]).
 if nargin < 3
     a0 = 1e-4;
     b0 = 1e-4;

chapter11/rndDirichlet.m

@@ -1,5 +1,5 @@
 function x = rndDirichlet(a)
 % Sampling from a Dirichlet distribution.
-% Written by Michael Chen ([email protected]).
+% Written by Mo Chen ([email protected]).
 x = gamrnd(a,1);
 x = x/sum(x);

chapter11/rndDiscrete.m

@@ -1,6 +1,6 @@
 function x = rndDiscrete(p, n)
 % Sampling from a discrete distribution (multinomial).
-% Written by Michael Chen ([email protected]).
+% Written by Mo Chen ([email protected]).
 if nargin == 1
     n = 1;
 end

chapter11/rndGauss.m

@@ -1,6 +1,6 @@
 function x = rndGauss(mu,Sigma,n)
 % Sampling from a Gaussian distribution.
-% Written by Michael Chen ([email protected]).
+% Written by Mo Chen ([email protected]).
 if nargin == 2
     n = 1;
 end

chapter12/dimFa.m

@@ -3,7 +3,7 @@
 %   X: d x n data matrix
 %   p: dimension of target space
 % Reference: Pattern Recognition and Machine Learning by Christopher M. Bishop 
-% Written by Michael Chen ([email protected]).
+% Written by Mo Chen ([email protected]).
 [d,n] = size(X);
 mu = mean(X,2);
 X = bsxfun(@minus,X,mu);

chapter12/dimPca.m

@@ -2,8 +2,7 @@
 % Perform standard PCA (spectral method).
 %   X: d x n data matrix
 %   p: dimension of target space (p>=1) or ratio (0<p<1)
-% Written by Michael Chen ([email protected]).
-
+% Written by Mo Chen ([email protected]).
 opts.disp = 0;
 opts.issym = 1;
 opts.isreal = 1;

chapter12/dimPcaEm.m

@@ -5,7 +5,7 @@
 % Reference: 
 %   Pattern Recognition and Machine Learning by Christopher M. Bishop 
 %   Probabilistic Principal Component Analysis by Michael E. Tipping & Christopher M. Bishop
-% Written by Michael Chen ([email protected]).
+% Written by Mo Chen ([email protected]).
 
 [m,n] = size(X);
 mu = mean(X,2);

chapter12/dimPcaLs.m

@@ -5,7 +5,7 @@
 % Reference: 
 %   Pattern Recognition and Machine Learning by Christopher M. Bishop 
 %   EM algorithms for PCA and SPCA by Sam Roweis 
-% Written by Michael Chen ([email protected]).
+% Written by Mo Chen ([email protected]).
 [d,n] = size(X);
 X = bsxfun(@minus,X,mean(X,2));
 W = rand(d,p); 

chapter12/dimPcaVb.m

@@ -5,8 +5,7 @@
 % Reference: 
 %   Pattern Recognition and Machine Learning by Christopher M. Bishop 
 %   Probabilistic Principal Component Analysis by Michael E. Tipping & Christopher M. Bishop
-% Written by Michael Chen ([email protected]).
-
+% Written by Mo Chen ([email protected]).
 [m,n] = size(X);
 if nargin < 3
     a0 = 1e-4;

chapter13/discreternd.m

@@ -1,6 +1,6 @@
 function x = discreternd(p, n)
 % Sampling from a discrete distribution (multinomial).
-% Written by Michael Chen ([email protected]).
+% Written by Mo Chen ([email protected]).
 if nargin == 1
     n = 1;
 end