写一段matlab基于bp神经网络的静脉识别代码

以下是基于bp神经网络的静脉识别的Matlab代码示例：

首先，加载数据集：

load vein_data.mat; % 加载静脉数据集
x_train = trainData'; % 训练数据
y_train = trainLabel'; % 训练标签
x_test = testData'; % 测试数据
y_test = testLabel'; % 测试标签

接下来，设置神经网络参数：

input_layer_size = size(x_train, 2); % 输入层大小
hidden_layer_size = 100; % 隐藏层大小
num_labels = size(y_train, 2); % 输出层大小
lambda = 0.1; % 正则化参数
max_iter = 200; % 最大迭代次数

然后，定义代价函数和梯度函数：

% 定义代价函数
function [J, grad] = costFunction(theta, input_layer_size, hidden_layer_size, num_labels, X, y, lambda)
m = size(X, 1); % 样本数量
Theta1 = reshape(theta(1:hidden_layer_size * (input_layer_size + 1)), hidden_layer_size, (input_layer_size + 1));
Theta2 = reshape(theta((1 + (hidden_layer_size * (input_layer_size + 1))):end), num_labels, (hidden_layer_size + 1));
J = 0;
Theta1_grad = zeros(size(Theta1));
Theta2_grad = zeros(size(Theta2));
X = [ones(m, 1) X];
a1 = X;
z2 = a1 * Theta1';
a2 = sigmoid(z2);
a2 = [ones(m, 1) a2];
z3 = a2 * Theta2';
h = sigmoid(z3);
y_matrix = eye(num_labels)(y,:);
J = (1/m) * sum(sum(-y_matrix .* log(h) - (1-y_matrix) .* log(1-h))) + (lambda/(2*m)) * (sum(sum(Theta1(:,2:end).^2)) + sum(sum(Theta2(:,2:end).^2)));
d3 = h - y_matrix;
d2 = (d3 * Theta2(:,2:end)) .* sigmoidGradient(z2);
Delta1 = d2' * a1;
Delta2 = d3' * a2;
Theta1_grad = Delta1./m + (lambda/m) * [zeros(size(Theta1, 1), 1) Theta1(:,2:end)];
Theta2_grad = Delta2./m + (lambda/m) * [zeros(size(Theta2, 1), 1) Theta2(:,2:end)];
grad = [Theta1_grad(:) ; Theta2_grad(:)];
end

% 定义Sigmoid函数
function g = sigmoid(z)
g = 1.0 ./ (1.0 + exp(-z));
end

% 定义Sigmoid函数的导数
function g = sigmoidGradient(z)
g = sigmoid(z) .* (1 - sigmoid(z));
end

接下来，训练神经网络：

initial_Theta1 = randInitializeWeights(input_layer_size, hidden_layer_size);
initial_Theta2 = randInitializeWeights(hidden_layer_size, num_labels);
initial_nn_params = [initial_Theta1(:) ; initial_Theta2(:)];
options = optimset('MaxIter', max_iter);
costFunction = @(p) costFunction(p, input_layer_size, hidden_layer_size, num_labels, x_train, y_train, lambda);
[nn_params, cost] = fmincg(costFunction, initial_nn_params, options);
Theta1 = reshape(nn_params(1:hidden_layer_size * (input_layer_size + 1)), hidden_layer_size, (input_layer_size + 1));
Theta2 = reshape(nn_params((1 + (hidden_layer_size * (input_layer_size + 1))):end), num_labels, (hidden_layer_size + 1));

最后，测试神经网络：

a1_test = [ones(size(x_test, 1), 1) x_test];
z2_test = a1_test * Theta1';
a2_test = sigmoid(z2_test);
a2_test = [ones(size(a2_test, 1), 1) a2_test];
z3_test = a2_test * Theta2';
h_test = sigmoid(z3_test);
[~, p_test] = max(h_test, [], 2);
accuracy = mean(double(p_test == y_test)) * 100;
fprintf('测试集准确率：%f%%', accuracy);

完整代码如下所示：

% 加载数据集
load vein_data.mat;
x_train = trainData';
y_train = trainLabel';
x_test = testData';
y_test = testLabel';

% 设置神经网络参数
input_layer_size = size(x_train, 2);
hidden_layer_size = 100;
num_labels = size(y_train, 2);
lambda = 0.1;
max_iter = 200;

% 定义代价函数和梯度函数
function [J, grad] = costFunction(theta, input_layer_size, hidden_layer_size, num_labels, X, y, lambda)
m = size(X, 1);
Theta1 = reshape(theta(1:hidden_layer_size * (input_layer_size + 1)), hidden_layer_size, (input_layer_size + 1));
Theta2 = reshape(theta((1 + (hidden_layer_size * (input_layer_size + 1))):end), num_labels, (hidden_layer_size + 1));
J = 0;
Theta1_grad = zeros(size(Theta1));
Theta2_grad = zeros(size(Theta2));
X = [ones(m, 1) X];
a1 = X;
z2 = a1 * Theta1';
a2 = sigmoid(z2);
a2 = [ones(m, 1) a2];
z3 = a2 * Theta2';
h = sigmoid(z3);
y_matrix = eye(num_labels)(y,:);
J = (1/m) * sum(sum(-y_matrix .* log(h) - (1-y_matrix) .* log(1-h))) + (lambda/(2*m)) * (sum(sum(Theta1(:,2:end).^2)) + sum(sum(Theta2(:,2:end).^2)));
d3 = h - y_matrix;
d2 = (d3 * Theta2(:,2:end)) .* sigmoidGradient(z2);
Delta1 = d2' * a1;
Delta2 = d3' * a2;
Theta1_grad = Delta1./m + (lambda/m) * [zeros(size(Theta1, 1), 1) Theta1(:,2:end)];
Theta2_grad = Delta2./m + (lambda/m) * [zeros(size(Theta2, 1), 1) Theta2(:,2:end)];
grad = [Theta1_grad(:) ; Theta2_grad(:)];
end

function g = sigmoid(z)
g = 1.0 ./ (1.0 + exp(-z));
end

function g = sigmoidGradient(z)
g = sigmoid(z) .* (1 - sigmoid(z));
end

% 训练神经网络
initial_Theta1 = randInitializeWeights(input_layer_size, hidden_layer_size);
initial_Theta2 = randInitializeWeights(hidden_layer_size, num_labels);
initial_nn_params = [initial_Theta1(:) ; initial_Theta2(:)];
options = optimset('MaxIter', max_iter);
costFunction = @(p) costFunction(p, input_layer_size, hidden_layer_size, num_labels, x_train, y_train, lambda);
[nn_params, cost] = fmincg(costFunction, initial_nn_params, options);
Theta1 = reshape(nn_params(1:hidden_layer_size * (input_layer_size + 1)), hidden_layer_size, (input_layer_size + 1));
Theta2 = reshape(nn_params((1 + (hidden_layer_size * (input_layer_size + 1))):end), num_labels, (hidden_layer_size + 1));

% 测试神经网络
a1_test = [ones(size(x_test, 1), 1) x_test];
z2_test = a1_test * Theta1';
a2_test = sigmoid(z2_test);
a2_test = [ones(size(a2_test, 1), 1) a2_test];
z3_test = a2_test * Theta2';
h_test = sigmoid(z3_test);
[~, p_test] = max(h_test, [], 2);
accuracy = mean(double(p_test == y_test)) * 100;
fprintf('测试集准确率：%f%%', accuracy);