Neural Networks Learning
这次试用的数据和上次是一样的数据。5000个training example，每一个代表一个数字的图像，图像是20x20的灰度图，400个像素的每个位置的灰度值组成了一个training example。
Model representation
这里我们使用如下的模型，共3层，其中第一层input有400个feature，另外还有bias unit。
说明一：Andrew已经为我们
准备好了神经网络的参数初始化值，只要我们load('ex4weights.mat'),即可得到该初始化值。
神经网络的训练是在该初始化的基础上开始训练的，如果
初始化值不合适，可能会收敛到局部最优解，这也是训练神经网络最容易犯的错误。
引申一下，DBN算法的思想是
先用DBN进行无监督训练，得到网络的初始参数，然后将
该参数来初始化神经网络，最后进行有监督的训练。原因就是，
DBN的无监督训练得到的初始参数来初始化神经网络，可以避免收敛到局部最优解。
Feedforward and cost function
现在我们需要计算cost function和gradient
这里是没有regularization的情况下的cost function公式，m表示m个training example，k表示k个output，这里我们有10个output，分别代表1-10
y的值需要从数字形式转化为向量形式，如下，下面的例子中第一列表示1，第二列表示2
Regularized cost function
下面是regularized cost function，其实就是在普通的cost function后面加上每个theta的平方，当然不要忘了系数。不过这里并不是所有的theta，所有跟bias unit相关的theta都要排除。
下面的代码是：nnCostFunction，只实现了正则化后的代价函数，若要实现非正则化的代价函数，则只需令 lambda=0。
function [J grad] = nnCostFunction(nn_params, ...
input_layer_size, ...
hidden_layer_size, ...
num_labels, ...
X, y, lambda)
%NNCOSTFUNCTION Implements the neural network cost function for a two layer
%neural network which performs classification
% [J grad] = NNCOSTFUNCTON(nn_params, hidden_layer_size, num_labels, ...
% X, y, lambda) computes the cost and gradient of the neural network. The
% parameters for the neural network are "unrolled" into the vector
% nn_params and need to be converted back into the weight matrices.
%
% The returned parameter grad should be a "unrolled" vector of the
% partial derivatives of the neural network.
%
% Reshape nn_params back into the parameters Theta1 and Theta2, the weight matrices
% for our 2 layer neural network
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));
% Setup some useful variables
m = size(X, 1);
% You need to return the following variables correctly
J = 0;
Theta1_grad = zeros(size(Theta1));
Theta2_grad = zeros(size(Theta2));
% ====================== YOUR CODE HERE ======================
% Instructions: You should complete the code by working through the
% following parts.
%
% Part 1: Feedforward the neural network and return the cost in the
% variable J. After implementing Part 1, you can verify that your
% cost function computation is correct by verifying the cost
% computed in ex4.m
%
% Part 2: Implement the backpropagation algorithm to compute the gradients
% Theta1_grad and Theta2_grad. You should return the partial derivatives of
% the cost function with respect to Theta1 and Theta2 in Theta1_grad and
% Theta2_grad, respectively. After implementing Part 2, you can check
% that your implementation is correct by running checkNNGradients
%
% Note: The vector y passed into the function is a vector of labels
% containing values from 1..K. You need to map this vector into a
% binary vector of 1's and 0's to be used with the neural network
% cost function.
%
% Hint: We recommend implementing backpropagation using a for-loop
% over the training examples if you are implementing it for the
% first time.
%
% Part 3: Implement regularization with the cost function and gradients.
%
% Hint: You can implement this around the code for
% backpropagation. That is, you can compute the gradients for
% the regularization separately and then add them to Theta1_grad
% and Theta2_grad from Part 2.
%
%%完全向量化版
a1 = [ones(m,1) X]; %5000*401
z2 = a1*Theta1'; %5000*25
a2 = [ones(size(z2,1),1) sigmoid(z2)]; %5000*(25+1)
z3 = a2*Theta2'; %5000*10
a3 = sigmoid(z3);
h=a3;
%-----------------Part 3: Compute Cost (Feedforward)------- -------------
Y=zeros(m,num_labels);
for i=1:num_labels
Y(:,i)=(y==i);
end
J = 1/m*sum(sum(-Y.*log(h)-(1-Y).*log(1-h)));
%--------------------------------------------------------------------------
%%正则化后
J = J + lambda/2/m*( sum(sum(Theta1(:,2:end).^2)) + sum(sum(Theta2(:,2:end).^2)) );
% compute delta
delta3=zeros(m, num_labels);
for k = 1 : num_labels
delta3(:,k) = a3(:,k) - (y==k); %5000*10
end
delta2 = delta3 * Theta2 .* [ones(size(z2,1),1) sigmoidGradient(z2)]; %5000*26
%compute Delta
Delta1 = delta2(:,2:end)' * a1; %25*401
Delta2 = delta3' * a2; %10*26
% compute Theta_grad
Theta1_grad = 1/m*Delta1;
Theta2_grad = 1/m*Delta2;
% 正则化grad
reg1 = lambda/m*Theta1;
reg2 = lambda/m*Theta2;
reg1(:,1) = 0;
reg2(:,1) = 0;
Theta1_grad = Theta1_grad + reg1;
Theta2_grad = Theta2_grad + reg2;
% -------------------------------------------------------------
% =========================================================================
% Unroll gradients
grad = [Theta1_grad(:) ; Theta2_grad(:)];
end
function [J grad] = nnCostFunction(nn_params, input_layer_size, hidden_layer_size, num_labels, X, y, lambda)
输入参数：nn_params-神经网络的参数
input_layer_size，hidden_layer_size, num_labels -描述神经网络的各层结点
X, y -神经网络的训练数据，输入层(X)和输出层(y)
正则化代价函数的表达式如下，可以看出J(θ)是关于训练数据（X,y），神经网络参数，还要正则化参数lambda的函数
所以nnCostFunction（）的输入参数包含以上的数据，自然就可以计算出，该数据下对应的代价值。
神经网络算法是通过反向传播来计算梯度的，即J(θ)的导数
在反向传播过程中，我们要计算除了input layer之外的每层的小 delta，也就是δ。参考下图，
计算过程分为下面5步，每个training example都循环进行1-4：
1 forward propagation计算每一层的函数z，和每一层的activation，不要忘记途中增加bias unit
2 计算第三层的δ，
3 计算第二层的δ，
4计算每一层的Δ，这里要搞清楚，δ代表当前层的所有unit个数，如果是hidden layer的话，还包括了bias unit。Δ的最终结果是当前层的theta的偏导数，和当前层的theta维数一致，通过δ计算Δ的时候，由于后层的δ中的bias unit并不是有当前层的theta计算到的，反向计算theta的偏导数，Δ的时候则要把其中的bias unit去掉
5 上面m个training example都循环计算完成之后，计算最后的偏导数
实际上这里的第一步，可以一次性使用矩阵操作计算完成所有m个训练数据的结果，但是后面的几步，还是要循环计算每个训练数据的结果
完成第一步：前向计算
------------------------------------------------------------------------------------------------------------------------------------------------------
1 forward propagation计算每一层的函数z，和每一层的activation，不要忘记途中增加bias unit
%%完全向量化版
a1 = [ones(m,1) X]; %5000*401
z2 = a1*Theta1'; %5000*25
a2 = [ones(size(z2,1),1) sigmoid(z2)]; %5000*(25+1)
z3 = a2*Theta2'; %5000*10
a3 = sigmoid(z3);
h=a3;
a1 = [ones(m,1) X]; %5000*401
z2 = a1*Theta1'; %5000*25
a2 = [ones(size(z2,1),1) sigmoid(z2)]; %5000*(25+1)
z3 = a2*Theta2'; %5000*10
a3 = sigmoid(z3);
h=a3;
---------------------------------------------------------------------------------------------------------------------------------------------------------------
计算正则化后的代价函数
J = 1/m*sum(sum(-Y.*log(h)-(1-Y).*log(1-h)));%--------------------------------------------------------------------------
%%正则化后
J = J + lambda/2/m*( sum(sum(Theta1(:,2:end).^2)) + sum(sum(Theta2(:,2:end).^2)) );
---------------------------------------------------------------------------------------------------------------------------------------------------------------
将训练数据的输出层转化为向量的形式
Y=zeros(m,num_labels);
for i=1:num_labels
Y(:,i)=(y==i);
end
-----------------------------------------------------------------------------------------------------------------------------------------------------------------
计算每一层的
delta3=zeros(m, num_labels);
for k = 1 : num_labels
delta3(:,k) = a3(:,k) - (y==k); %5000*10
end
delta2 = delta3 * Theta2 .* [ones(size(z2,1),1) sigmoidGradient(z2)]; %5000*26
%compute Delta
Delta1 = delta2(:,2:end)' * a1; %25*401
Delta2 = delta3' * a2; %10*26
% compute Theta_grad
Theta1_grad = 1/m*Delta1;
Theta2_grad = 1/m*Delta2;
-------------------------------------------------------------------------------------------------------------------------------------------------
Regularized Neural Networks
计算的时候，所有和bias项对应的theta都不需要做regularization
% 正则化grad
reg1 = lambda/m*Theta1;
reg2 = lambda/m*Theta2;
reg1(:,1) = 0;
reg2(:,1) = 0;
Theta1_grad = Theta1_grad + reg1;
Theta2_grad = Theta2_grad + reg2;
---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Random initialization
为了打破symmetry breaking，我们要在选取初始theta的时候，每次选择某一层的theta，然后让每个这一层的theta都随机选取，范围 [-ξ，ξ]
有如下经验公式可用来选择合适的theta范围
确定了范围之后，下面是随机选择的过程
% Randomly initialize the weights to small values
epsilon init = 0.12;
W = rand(L out, 1 + L in) * 2 * epsilon init − epsilon init;Gradient checking
梯度校验法是本周课程中非常值得学习的技巧。首先一个问题，如何判断我们神经网络中计算梯度（代价函数的导数是否是正确的呢）。
首先，明确一点，我们使用反向传播算法来计算梯度的，并不是直接使用代价函数的解析式求导之后来计算，因为该代价函数求导之后找不到解析式。回想我们在函数 nnCostFunction()中计算出了什么？算出了在神经网络参数nn_params，和对应训练数据下的代价函数值，和对应的导数值（梯度），在该函数中，我们是用反向传播算法来计算梯度的。代价函数可以是看出关于神经网络参数θ的函数J(θ)。
所以，我们可以考虑利用数值计算的方法来计算函数J(θ)的导数，来与利用反向传播算法计算得到的梯度进行比较。如果，代价函数计算正确，反向传播计算梯度的方法正确，那么数值计算得到的梯度和反向传播计算得到的梯度，应该是非常相近的。
先把神经网络的各层参数Θ(1), Θ(2)，整合在一个向量里面θ，那么代价函数J(θ)就必须对该向量θ的每一个分量（θ1，θ2，θ3，.....θn）求偏导，得到关于每一个θ分量的梯度。
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
function checkNNGradients(lambda)
%CHECKNNGRADIENTS Creates a small neural network to check the
%backpropagation gradients
% CHECKNNGRADIENTS(lambda) Creates a small neural network to check the
% backpropagation gradients, it will output the analytical gradients
% produced by your backprop code and the numerical gradients (computed
% using computeNumericalGradient). These two gradient computations should
% result in very similar values.
%
if ~exist('lambda', 'var') || isempty(lambda)
lambda = 0;
end
input_layer_size = 3;
hidden_layer_size = 5;
num_labels = 3;
m = 5;
% We generate some 'random' test data
Theta1 = debugInitializeWeights(hidden_layer_size, input_layer_size);
Theta2 = debugInitializeWeights(num_labels, hidden_layer_size);
% Reusing debugInitializeWeights to generate X
X = debugInitializeWeights(m, input_layer_size - 1);
y = 1 + mod(1:m, num_labels)';
% Unroll parameters
nn_params = [Theta1(:) ; Theta2(:)];
% Short hand for cost function
costFunc = @(p) nnCostFunction(p, input_layer_size, hidden_layer_size, ...
num_labels, X, y, lambda);
[cost, grad] = costFunc(nn_params);
numgrad = computeNumericalGradient(costFunc, nn_params);
% Visually examine the two gradient computations. The two columns
% you get should be very similar.
disp([numgrad grad]);
fprintf(['The above two columns you get should be very similar.\n' ...
'(Left-Your Numerical Gradient, Right-Analytical Gradient)\n\n']);
% Evaluate the norm of the difference between two solutions.
% If you have a correct implementation, and assuming you used EPSILON = 0.0001
% in computeNumericalGradient.m, then diff below should be less than 1e-9
diff = norm(numgrad-grad)/norm(numgrad+grad);
fprintf(['If your backpropagation implementation is correct, then \n' ...
'the relative difference will be small (less than 1e-9). \n' ...
'\nRelative Difference: %g\n'], diff);
end<span style="color:#ff0000;">
</span>
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
%获取nnCostFunction的函数指针，传人相应参数
costFunc = @(p) nnCostFunction(p, input_layer_size, hidden_layer_size, ...
num_labels, X, y, lambda);
%利用nnCostFunction计算相应参数下的梯度值
[cost, grad] = costFunc(nn_params);
num_labels, X, y, lambda);
%利用nnCostFunction计算相应参数下的梯度值
[cost, grad] = costFunc(nn_params);
%利用数值计算computeNumericalGradient来计算梯度值
numgrad = computeNumericalGradient(costFunc, nn_params);
numgrad = computeNumericalGradient(costFunc, nn_params);
% Visually examine the two gradient computations. The two columns
% you get should be very similar.
disp([numgrad grad]);
从函数的调用可知 numgrad = computeNumericalGradient(costFunc, nn_params);J就是nnCostFunction的函数指针
function numgrad = computeNumericalGradient(J, theta)
%COMPUTENUMERICALGRADIENT Computes the gradient using "finite differences"
%and gives us a numerical estimate of the gradient.
% numgrad = COMPUTENUMERICALGRADIENT(J, theta) computes the numerical
% gradient of the function J around theta. Calling y = J(theta) should
% return the function value at theta.
% Notes: The following code implements numerical gradient checking, and
% returns the numerical gradient.It sets numgrad(i) to (a numerical
% approximation of) the partial derivative of J with respect to the
% i-th input argument, evaluated at theta. (i.e., numgrad(i) should
% be the (approximately) the partial derivative of J with respect
% to theta(i).)
%
numgrad = zeros(size(theta));
perturb = zeros(size(theta));
e = 1e-4;
for p = 1:numel(theta)
% Set perturbation vector
perturb(p) = e;
loss1 = J(theta - perturb);
loss2 = J(theta + perturb);
% Compute Numerical Gradient
numgrad(p) = (loss2 - loss1) / (2*e);
perturb(p) = 0;
end
end
for p = 1:numel(theta)
% Set perturbation vector
perturb(p) = e;
%J就是nnCostFunction的函数指针,计算θ=theta - perturb，下的代价函数
loss1 = J(theta - perturb);
%计算θ=theta+perturb，下的代价函数
loss2 = J(theta + perturb);
% Compute Numerical Gradient
%数值计算公式求梯度。
numgrad(p) = (loss2 - loss1) / (2*e);
perturb(p) = 0;
end
Backpropagation
Sigmoid gradient
sigmoid gradient反向传播计算gradient的公式的一部分
function g = sigmoidGradient(z)
%SIGMOIDGRADIENT returns the gradient of the sigmoid function
%evaluated at z
% g = SIGMOIDGRADIENT(z) computes the gradient of the sigmoid function
% evaluated at z. This should work regardless if z is a matrix or a
% vector. In particular, if z is a vector or matrix, you should return
% the gradient for each element.
g = zeros(size(z));
% ====================== YOUR CODE HERE ======================
% Instructions: Compute the gradient of the sigmoid function evaluated at
% each value of z (z can be a matrix, vector or scalar).
g = sigmoid(z).*(1-sigmoid(z));
% =============================================================
end
3193
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
