请写出基于matlab的Fisher线性判别函数的模板

以下是基于Matlab的Fisher线性判别函数的模板：

function [W, b] = fisher_ld(X, Y)
% Input:
% X: n-by-d data matrix, each row is a data point and each column is a feature
% Y: n-by-1 label vector
% Output:
% W: d-by-1 weight vector of Fisher linear discriminant
% b: scalar bias

% Compute the mean of each class
class_labels = unique(Y);
num_classes = length(class_labels);
means = zeros(num_classes, size(X, 2));
for i = 1:num_classes
    means(i, :) = mean(X(Y==class_labels(i), :), 1);
end

% Compute the within-class scatter matrix
Sw = zeros(size(X, 2));
for i = 1:num_classes
    Xi = X(Y==class_labels(i), :);
    Si = (Xi - means(i, :))' * (Xi - means(i, :));
    Sw = Sw + Si;
end

% Compute the between-class scatter matrix
Sb = zeros(size(X, 2));
for i = 1:num_classes
    Ni = sum(Y==class_labels(i));
    mi = means(i, :);
    Sb = Sb + Ni * (mi - mean(X, 1))' * (mi - mean(X, 1));
end

% Compute the Fisher linear discriminant
[W, D] = eig(Sb, Sw);
[~, ind] = max(abs(diag(D)));
W = real(W(:, ind));
W = W / norm(W, 2);

% Compute the bias
b = -mean(X * W(Y==class_labels(1), :)' + X * W(Y==class_labels(2), :)', 1) / 2;
end

该函数将数据矩阵和标签向量作为输入，并返回Fisher线性判别函数的权重向量和偏置项。函数首先计算每个类别的均值，然后计算类内散布矩阵和类间散布矩阵。接下来，函数通过求解广义特征值问题来计算Fisher线性判别函数的权重向量。最后，函数计算偏置项，使得Fisher线性判别函数在两个类别的中间点处为零。