extracted demos from ch01~08

chapter01/demo.m → demo/ch01/info_demo.m

@@ -7,9 +7,6 @@
 x = ceil(k*rand(1,n));
 y = ceil(k*rand(1,n));
 
-% x = randi(k,1,n);  % need statistics toolbox
-% y = randi(k,1,n);
-
 %% Entropy H(x), H(y)
 Hx = entropy(x);
 Hy = entropy(y);

demo/ch03/linRegEm_demo.m

@@ -0,0 +1,14 @@
+% demos for ch03
+clear; close all;
+d = 1;
+n = 200;
+[x,t] = linRnd(d,n);
+%% Empirical Bayesian linear regression via EM
+[model,llh] = linRegEm(x,t);
+plot(llh);
+[y,sigma] = linRegPred(model,x,t);
+figure
+plotCurveBar(x,y,sigma);
+hold on;
+plot(x,t,'o');
+hold off;

demo/ch03/linRegFp_demo.m

@@ -0,0 +1,16 @@
+% demos for ch03
+clear; close all;
+d = 1;
+n = 200;
+[x,t] = linRnd(d,n);
+%%  Empirical Bayesian linear regression via Mackay fix point iteration method
+[model,llh] = linRegFp(x,t);
+plot(llh);
+[y,sigma] = linRegPred(model,x,t);
+figure
+plotCurveBar(x,y,sigma);
+hold on;
+plot(x,t,'o');
+hold off;
+%%
+

demo/ch03/linReg_demo.m

@@ -0,0 +1,12 @@
+% demos for ch03
+clear; close all;
+d = 1;
+n = 200;
+[x,t] = linRnd(d,n);
+%% Linear regression
+model = linReg(x,t);
+[y,sigma] = linRegPred(model,x,t);
+plotCurveBar( x, y, sigma );
+hold on;
+plot(x,t,'o');
+hold off;

demo/ch04/logitBin_demo.m

@@ -0,0 +1,11 @@
+% demos for ch04
+
+%% Logistic logistic regression for binary classification
+clear; close all;
+k = 2;
+n = 1000;
+[X,t] = kmeansRnd(2,k,n);
+[model, llh] = logitBin(X,t-1);
+plot(llh);
+y = logitBinPred(model,X)+1;
+binPlot(model,X,y)

demo/ch04/logitMn_demo.m

@@ -0,0 +1,8 @@
+%% Logistic logistic regression for multiclass classification
+clear
+k = 3;
+n = 1000;
+[X,t] = kmeansRnd(2,k,n);
+[model, llh] = logitMn(X,t);
+y = logitMnPred(model,X);
+plotClass(X,y)

demo/ch06/knCenter_demo.m

@@ -0,0 +1,9 @@
+%% demo for knCenter
+clear; close all;
+kn = @knGauss;
+X=rand(2,100);
+X1=rand(2,10);
+X2=rand(2,5);
+
+maxdiff(knCenter(kn,X,X1),diag(knCenter(kn,X,X1,X1))')
+maxdiff(knCenter(kn,X),knCenter(kn,X,X,X))

chapter06/demo.m → demo/ch06/knLin_demo.m

@@ -1,80 +1,61 @@
-demos for ch06
-
-
-%% Kernel regression with gaussian kernel
-clear; close all;
-n = 100;
-x = linspace(0,2*pi,n);   % test data
-t = sin(x)+rand(1,n)/2;
-model = knReg(x,t,1e-4,@knGauss);
-[y,s] = knRegPred(model,x);
-plotCurveBar(x,y,s);
-hold on;
-plot(x,t,'o');
-hold off;
-%% Kernel regression with linear kernel is EQUIVALENT to linear regression
-lambda = 1e-4;
-model_kn = knReg(x,t,lambda,@knLin);
-model_lin = linReg(x,t,lambda);
-
-idx = 1:2:n;
-xt = x(:,idx);
-tt = t(idx);
-
-[y_kn, sigma_kn,p_kn] = knRegPred(model_kn,xt,tt);
-[y_lin, sigma_lin,p_lin] = linRegPred(model_lin,xt,tt);
-
-maxdiff(y_kn,y_lin)
-maxdiff(sigma_kn,sigma_lin)
-maxdiff(p_kn,p_lin)
-%% Kernel kmeans with linear kernel is EQUIVALENT to kmeans
-clear; close all;
-d = 2;
-k = 3;
-n = 500;
-[X,y] = kmeansRnd(d,k,n);
-init = ceil(k*rand(1,n));
-[y_kn,en_kn,model_kn] = knKmeans(X,init,@knLin);
-[y_lin,en_lin,model_lin] = kmeans(X,init);
-
-idx = 1:2:n;
-Xt = X(:,idx);
-
-[t_kn,ent_kn] = knKmeansPred(model_kn, Xt);
-[t_lin,ent_lin] = kmeansPred(model_lin, Xt);
-
-maxdiff(y_kn,y_lin)
-maxdiff(en_kn,en_lin)
-
-maxdiff(t_kn,t_lin)
-maxdiff(ent_kn,ent_lin)
-%% Kernel PCA with linear kernel is EQUIVALENT TO PCA
-clear; close all;
-d = 10;
-q = 2;
-n = 500;
-X = randn(d,n);
-
-
-model_kn = knPca(X,q,@knLin);
-idx = 1:2:n;
-Xt = X(:,idx);
-
-Y_kn = knPcaPred(model_kn,Xt);
-
-[U,L,mu,mse] = pca(X,q);
-Y_lin = U'*bsxfun(@minus,Xt,mu);   % projection
-
-
-R = Y_lin/Y_kn;    % the results are equivalent up to a rotation.
-maxdiff(R*R', eye(q))
-
-%% demo for knCenter
-clear; close all;
-kn = @knGauss;
-X=rand(2,100);
-X1=rand(2,10);
-X2=rand(2,5);
-
-maxdiff(knCenter(kn,X,X1),diag(knCenter(kn,X,X1,X1))')
-maxdiff(knCenter(kn,X),knCenter(kn,X,X,X))
+%% Kernel regression with linear kernel is EQUIVALENT to linear regression
+clear; close all;
+n = 100;
+x = linspace(0,2*pi,n);   % test data
+t = sin(x)+rand(1,n)/2;
+
+lambda = 1e-4;
+model_kn = knReg(x,t,lambda,@knLin);
+model_lin = linReg(x,t,lambda);
+
+idx = 1:2:n;
+xt = x(:,idx);
+tt = t(idx);
+
+[y_kn, sigma_kn,p_kn] = knRegPred(model_kn,xt,tt);
+[y_lin, sigma_lin,p_lin] = linRegPred(model_lin,xt,tt);
+
+maxdiff(y_kn,y_lin)
+maxdiff(sigma_kn,sigma_lin)
+maxdiff(p_kn,p_lin)
+%% Kernel kmeans with linear kernel is EQUIVALENT to kmeans
+clear; close all;
+d = 2;
+k = 3;
+n = 500;
+[X,y] = kmeansRnd(d,k,n);
+init = ceil(k*rand(1,n));
+[y_kn,en_kn,model_kn] = knKmeans(X,init,@knLin);
+[y_lin,en_lin,model_lin] = kmeans(X,init);
+
+idx = 1:2:n;
+Xt = X(:,idx);
+
+[t_kn,ent_kn] = knKmeansPred(model_kn, Xt);
+[t_lin,ent_lin] = kmeansPred(model_lin, Xt);
+
+maxdiff(y_kn,y_lin)
+maxdiff(en_kn,en_lin)
+
+maxdiff(t_kn,t_lin)
+maxdiff(ent_kn,ent_lin)
+%% Kernel PCA with linear kernel is EQUIVALENT TO PCA
+clear; close all;
+d = 10;
+q = 2;
+n = 500;
+X = randn(d,n);
+
+
+model_kn = knPca(X,q,@knLin);
+idx = 1:2:n;
+Xt = X(:,idx);
+
+Y_kn = knPcaPred(model_kn,Xt);
+
+[U,L,mu,mse] = pca(X,q);
+Y_lin = U'*bsxfun(@minus,Xt,mu);   % projection
+
+
+R = Y_lin/Y_kn;    % the results are equivalent up to a rotation.
+maxdiff(R*R', eye(q))

demo/ch06/knReg_demo.m

@@ -0,0 +1,14 @@
+% demos for ch06
+
+
+%% Kernel regression with gaussian kernel
+clear; close all;
+n = 100;
+x = linspace(0,2*pi,n);   % test data
+t = sin(x)+rand(1,n)/2;
+model = knReg(x,t,1e-4,@knGauss);
+[y,s] = knRegPred(model,x);
+plotCurveBar(x,y,s);
+hold on;
+plot(x,t,'o');
+hold off;

demo/ch07/rvmBinEm_demo.m

@@ -0,0 +1,12 @@
+%% RVM for classification
+clear; close all
+k = 2;
+d = 2;
+n = 1000;
+[X,t] = kmeansRnd(d,k,n);
+
+[model, llh] = rvmBinEm(X,t-1);
+plot(llh);
+y = rvmBinPred(model,X)+1;
+figure;
+binPlot(model,X,y);

chapter07/demo2.m → demo/ch07/rvmBinFp_demo.m

@@ -4,7 +4,6 @@
 d = 2;
 n = 1000;
 [X,t] = kmeansRnd(d,k,n);
-[x1,x2] = meshgrid(linspace(min(X(1,:)),max(X(1,:)),n), linspace(min(X(2,:)),max(X(2,:)),n));
 
 [model, llh] = rvmBinFp(X,t-1);
 plot(llh);

demo/ch07/rvmRegEm_demo.m

@@ -0,0 +1,18 @@
+%% regression
+d = 100;
+beta = 1e-1;
+X = rand(1,d);
+w = randn;
+b = randn;
+t = w'*X+b+beta*randn(1,d);
+x = linspace(min(X),max(X),d);   % test data
+
+%% RVM regression by EM
+[model,llh] = rvmRegEm(X,t);
+plot(llh);
+[y, sigma] = linRegPred(model,x,t);
+figure
+plotCurveBar(x,y,sigma);
+hold on;
+plot(X,t,'o');
+hold off