svm-matlab.rar_SVM matlab_d star_d-star算法_train_train_features

[matlab]支持向量机(SVM)用于分类的算法实现

function [D, a_star] = SVM(train_features, train_targets, params, region)

% Classify using (a very simple implementation of) the support vector machine algorithm

%

% Inputs:

% features- Train features

% targets - Train targets

% params - [kernel, kernel parameter, solver type, Slack]

% Kernel can be one of: Gauss, RBF (Same as Gauss), Poly, Sigmoid, or Linear

% Solver type can be one of: Perceptron, Quadprog

% region - Decision region vector: [-x x -y y number_of_points]

%

% Outputs

% D - Decision sufrace

% a - SVM coeficients

%

% Note: The number of support vectors found will usually be larger than is actually

% needed because both solvers are approximate.

[Dim, Nf] = size(train_features);

y = zeros(Nf);

switch kernel,

case {'Gauss','RBF'},

for i = 1:Nf,

y(:,i) = exp(-sum((train_features-train_features(:,i)*ones(1,Nf)).^2)'/(2*ker_param^2));

end

case {'Poly', 'Linear'}

if strcmp(kernel, 'Linear')

ker_param = 1;

end

y(:,i) = (train_features'*train_features(:,i) + 1).^ker_param;

end

case 'Sigmoid'

if (length(ker_param) ~= 2)

error('This kernel needs two parameters to operate!')

end

%Find the SVM coefficients

switch solver

case 'Quadprog'

%Quadratic programming

if ~isfinite(slack)

alpha_star = quadprog((z'*z).*(y'*y), -ones(1, Nf), [], [], z, 0, 0)';

else

alpha_star = quadprog((z'*z).*(y'*y), -ones(1, Nf), [], [], z, 0, 0, slack)';

a_star = (alpha_star.*z)*y';

end

case 'Perceptron'

max_iter = 1e5;

iter = 0;

rate = 0.01;

xi = ones(1,Nf)/Nf*slack;