def compute_cost_with_regularization(A3, Y, parameters, lambd):
"""
Implement the cost function with L2 regularization. See formula (2) above.
Arguments:
A3 -- post-activation, output of forward propagation, of shape (output size, number of examples)
Y -- "true" labels vector, of shape (output size, number of examples)
parameters -- python dictionary containing parameters of the model
Returns:
cost - value of the regularized loss function (formula (2))
"""
m = Y.shape[1]
W1 = parameters["W1"]
W2 = parameters["W2"]
W3 = parameters["W3"]
cross_entropy_cost = compute_cost(A3, Y) # This gives you the cross-entropy part of the cost
### START CODE HERE ### (approx. 1 line)
L2_regularization_cost = 1.0 / m * lambd / 2 * (np.sum(np.square(W1)) + np.sum(np.square(W2)) + np.sum(np.square(W3)))
### END CODER HERE ###
cost = cross_entropy_cost + L2_regularization_cost
return cost
def model(X,
Y,
learning_rate=0.3,
num_iterations=30000,
print_cost=True,
lambd=0,
keep_prob=1):
grads = {}
costs = []
m = X.shape[1]
layers_dims = [X.shape[0], 20, 3, 1]
parameters = initialize_parameters(layers_dims)
for i in range(0, num_iterations):
if keep_prob == 1:
a3, cache = forward_propagation(X, parameters)
elif keep_prob < 1:
a3, cache = forward_propagation_with_dropout_test_case(
X, parameters, keep_prob)
if lambd == 0:
cost = compute_cost(a3, Y)
else:
cost = compute_cost_with_regularization(a3, Y, parameters, lambd)
assert (lambd == 0 or keep_prob == 1)
if lambd == 0 and keep_prob == 1:
grads = backward_propagation(X, Y, cache)
elif lambd != 0:
grads = backward_propagation_with_regularization(
X, Y, cache, lambd)
elif keep_prob < 1:
backward_propagation_with_dropout(X, Y, cache, keep_prob)
parameters = update_parameters(parameters, grads, learning_rate)
if print_cost and i % 10000 == 0:
print("Cost after iteration {}: {}".format(i, cost))
if print_cost and i % 1000 == 0:
costs.append(cost)
plt.plot(costs)
plt.ylabel('cost')
plt.xlabel('iterations (x1,000)')
plt.title("Learning rate =" + str(learning_rate))
plt.show()
return parameters
def compute_cost_with_regularization(A3,Y,parameters,lambd):
"""
L2 正则化成本计算
参数:
A3 -正向传播的输出结果,维度(输出节点数量,训练/测试数量)
Y -标签向量 维度(输出节点数量,训练/测试数量)
parameters -模型学习后的参数字典
lambd -正则化超参数
返回:
cost -正则化损失的值
"""
m=Y.shape[1]
W1=parameters["W1"]
W2=parameters["W2"]
W3=parameters["W3"]
cross_entropy_cost=reg_utils.compute_cost(A3,Y)
L2_regularization_cost=lambd*(np.sum(np.square(W1))+np.sum(np.square(W2))+np.sum(np.square(W3)))/(2*m)
cost=cross_entropy_cost+L2_regularization_cost
return cost
def compute_cost_with_regularization(A3, Y, parameters, lambd):
"""
Implement the cost function with L2 regularization.
Arguments:
A3 -- post-activation, output of forward propagation, of shape (output size, number of examples)
Y -- "true" labels vector, of shape (output size, number of examples)
parameters -- python dictionary containing parameters of the model
Returns:
cost - value of the regularized loss function
"""
m = Y.shape[1]
cost = compute_cost(A3, Y) # without Regularization term
L = len(parameters) // 2
W = 0
for i in range(L):
W += np.sum(parameters["W" + str(i + 1)]**2)
cost += (lambd * W) / (2 * m)
return cost
def compute_cost_with_regularization(A3,Y,parameters,lambd):
'''
实现L2正则化计算成本
:param A3: 正向传播的结果,维度为(输出节点数量,训练/测试的数量)
:param Y: 标签向量,与数据一一对应,维度为(输出节点数量,驯良/测试的数量)
:param parameters: 包含模型学习后的参数的字典
:param lambd:
:return:
cost - 使用公式2计算出来的正则化损失函值
'''
m=Y.shape[1]
W1 = parameters['W1']
W2 = parameters['W2']
W3 = parameters['W3']
cross_entropy_cost = reg_utils.compute_cost(A3,Y)
L2_regularization_cost = lambd*(np.sum(np.square(W1))+np.sum(np.square(W2))+np.sum(np.square(W3)))/(m*2)
cost = cross_entropy_cost + L2_regularization_cost
return cost
def compute_cost_with_regularization(A3, Y, parameters, lambd):
"""
Implement the cost function with L2 regularization. See formula (2) above.
Arguments:
A3 -- post-activation, output of forward propagation, of shape (output size, number of examples)
Y -- "true" labels vector, of shape (output size, number of examples)
parameters -- python dictionary containing parameters of the model
Returns:
cost - value of the regularized loss function (formula (2))
"""
m = Y.shape[1]
W1 = parameters["W1"]
W2 = parameters["W2"]
W3 = parameters["W3"]
cross_entropy_cost = compute_cost(A3, Y) # This gives you the cross-entropy part of the cost
L2_regularization_cost = (np.sum(np.square(W1))+np.sum(np.square(W2))+np.sum(np.square(W3)))*lambd/2./m
cost = cross_entropy_cost + L2_regularization_cost
return cost
def compute_cost_with_regularization(A3, Y, parameters, lambd):
'''
实现L2正则化计算成本
:param A3: 正向传播的结果,维度为(输出节点数量,训练/测试的数量)
:param Y: 标签向量,与数据一一对应,维度为(输出节点数量,驯良/测试的数量)
:param parameters: 包含模型学习后的参数的字典
:param lambd:
:return:
cost - 使用公式2计算出来的正则化损失函值
'''
m = Y.shape[1]
W1 = parameters['W1']
W2 = parameters['W2']
W3 = parameters['W3']
cross_entropy_cost = reg_utils.compute_cost(A3, Y)
L2_regularization_cost = lambd * (np.sum(np.square(W1)) + np.sum(
np.square(W2)) + np.sum(np.square(W3))) / (m * 2)
cost = cross_entropy_cost + L2_regularization_cost
return cost
def compute_cost_with_regularization(A3, Y, parameters, lambd):
"""
实现公式2的L2正则化计算成本
参数：
A3 - 正向传播的输出结果，维度为（输出节点数量，训练/测试的数量）
Y - 标签向量，与数据一一对应，维度为(输出节点数量，训练/测试的数量)
parameters - 包含模型学习后的参数的字典
返回：
cost - 使用公式2计算出来的正则化损失的值
"""
m = Y.shape[1]
W1 = parameters["W1"]
W2 = parameters["W2"]
W3 = parameters["W3"]
cross_entropy_cost = reg_utils.compute_cost(A3, Y)
L2_regularization_cost = lambd * (np.sum(np.square(W1)) + np.sum(
np.square(W2)) + np.sum(np.square(W3))) / (2 * m)
cost = cross_entropy_cost + L2_regularization_cost
return cost
def compute_cost_with_regularization(A3, Y, parameters, lambd):
"""
Implement the cost function with L2 regularization. See formula (2) above.
Arguments:
A3 -- post-activation, output of forward propagation, of shape (output size, number of examples)
Y -- "true" labels vector, of shape (output size, number of examples)
parameters -- python dictionary containing parameters of the model
Returns:
cost - value of the regularized loss function (formula (2))
"""
m = Y.shape[1]
W1 = parameters["W1"]
W2 = parameters["W2"]
W3 = parameters["W3"]
# Modify the cost and observe the consequences
cross_entropy_cost = compute_cost(A3, Y) # This gives you the cross-entropy part of the cost
L2_regularization_cost = lambd * (np.sum(np.square(W1)) + np.sum(np.square(W2)) + np.sum(np.square(W3))) / (2 * m) # do this for the weights as well to sum up the three terms
cost = cross_entropy_cost + L2_regularization_cost
return cost
def model(X, Y, learning_rate = 0.3, num_iterations = 30000, print_cost = True, lambd = 0, keep_prob = 1):
"""
Implements a three-layer neural network: LINEAR->RELU->LINEAR->RELU->LINEAR->SIGMOID.
Arguments:
X -- input data, of shape (input size, number of examples)
Y -- true "label" vector (1 for blue dot / 0 for red dot), of shape (output size, number of examples)
learning_rate -- learning rate of the optimization
num_iterations -- number of iterations of the optimization loop
print_cost -- If True, print the cost every 10000 iterations
lambd -- regularization hyperparameter, scalar
keep_prob - probability of keeping a neuron active during drop-out, scalar.
Returns:
parameters -- parameters learned by the model. They can then be used to predict.
"""
grads = {}
costs = [] # to keep track of the cost
m = X.shape[1] # number of examples
layers_dims = [X.shape[0], 20, 3, 1]
# Initialize parameters dictionary.
parameters = initialize_parameters(layers_dims)
# Loop (gradient descent)
for i in range(0, num_iterations):
# Forward propagation: LINEAR -> RELU -> LINEAR -> RELU -> LINEAR -> SIGMOID.
if keep_prob == 1:
a3, cache = forward_propagation(X, parameters)
elif keep_prob < 1:
a3, cache = forward_propagation_with_dropout(X, parameters, keep_prob)
# Cost function
if lambd == 0:
cost = compute_cost(a3, Y)
else:
cost = compute_cost_with_regularization(a3, Y, parameters, lambd)
# Backward propagation.
assert(lambd==0 or keep_prob==1) # it is possible to use both L2 regularization and dropout,
# but this assignment will only explore one at a time
if lambd == 0 and keep_prob == 1:
grads = backward_propagation(X, Y, cache)
elif lambd != 0:
grads = backward_propagation_with_regularization(X, Y, cache, lambd)
elif keep_prob < 1:
grads = backward_propagation_with_dropout(X, Y, cache, keep_prob)
# Update parameters.
parameters = update_parameters(parameters, grads, learning_rate)
# Print the loss every 10000 iterations
if print_cost and i % 10000 == 0:
print("Cost after iteration {}: {}".format(i, cost))
if print_cost and i % 1000 == 0:
costs.append(cost)
# plot the cost
plt.plot(costs)
plt.ylabel('cost')
plt.xlabel('iterations (x1,000)')
plt.title("Learning rate =" + str(learning_rate))
plt.show()
return parameters
def model(X,
Y,
learning_rate=0.3,
num_iterations=30000,
print_cost=True,
lambd=0,
keep_prob=1):
grads = {}
costs = []
m = X.shape[1]
layer_dims = [X.shape[0], 20, 3, 1]
parameters = initialize_parameters(layer_dims)
for i in range(num_iterations):
if keep_prob == 1:
A3, cache = forward_propagation(X, parameters)
elif keep_prob < 1:
A3, cache = forward_propagation_with_dropout(
X, parameters, keep_prob)
if lambd == 0:
cost = compute_cost(A3, Y)
else:
cost = compute_cost_with_regulations(A3, Y, parameters, lambd)
assert (lambd == 0 or keep_prob == 1)
if lambd == 0 and keep_prob == 1:
grads = backward_propagation(X, Y, cache)
elif lambd != 0:
grads = backward_propagation_with_regularization(
X, Y, cache, lambd)
elif keep_prob < 1:
grads = backward_propagation_with_dropout(X, Y, cache, keep_prob)
parameters = update_parameters(parameters, grads, learning_rate)
if print_cost and i % 1000 == 0:
print("Cost After iterations {} : {}".format(i, cost))
costs.append(cost)
plt.plot(costs)
plt.title("Learning rate = " + str(learning_rate))
plt.ylabel("Cost")
plt.xlabel("iterations(x1,000)")
#plt.show()
plt.savefig('/home/guanlingh/local2server/homework6/image/h6-2.jpg')
return parameters
def model(X, Y, learning_rate = 0.3, num_iterations = 30000, print_cost = True, is_plot = True, lambd = 0, keep_prob = 1):
"""
实现一个三层的神经网络：ReLU->ReLU->Sigmoid
:param lambd:正则化的超参数
:param keep_prob:随机删除结点的概率
:return:
parameters - 学习后的参数
"""
parameter = {}
grads = {}
costs = []
m = X.shape[1]
layers_dims = [X.shape[0], 20, 3, 1]
# 初始化参数
parameters = reg_utils.initialize_parameters(layers_dims)
for l in range(num_iterations):
# 前向传播
# 是否随机删除结点
if keep_prob == 1:
# 前向传播
a3, cache = reg_utils.forward_propagation(X, parameters)
elif keep_prob < 1:
a3, cache = forward_propagation_with_dropout(X, parameters, keep_prob)
else:
print("keep_prob参数错误!程序退出")
exit()
# 计算成本
# 是否使用正则化
if lambd == 0:
cost = reg_utils.compute_cost(a3, Y)
else:
cost = compute_cost_with_regularization(a3, Y, parameters, lambd)
if l % 1000 == 0:
costs.append(cost)
if print_cost:
print(f'第{l}次迭代,成本值为:{cost}')
# 反向传播
# 可以同时使用L2正则化和随机删除节点,但本次实验不同时使用
assert(lambd == 0 or keep_prob == 1)
if lambd == 0 and keep_prob == 1:
# 不使用L2正则化和随机删除节点
grads = reg_utils.backward_propagation(X, Y, cache)
elif lambd != 0:
# 使用L2正则化,不使用dropout
grads = backward_propagation_with_regularization(X, Y, cache, lambd)
elif keep_prob < 1:
# 使用dropout,不使用L2正则化
grads = backward_propagation_with_dropout(X, Y, cache, keep_prob)
# 更新参数
parameters = reg_utils.update_parameters(parameters, grads, learning_rate)
if is_plot:
plt.plot(costs)
plt.ylabel('cost')
plt.xlabel('iterations (x1,000)')
plt.title("Learning rate =" + str(learning_rate))
plt.show()
return parameters
def model(X,
Y,
learning_rate=0.3,
num_iterations=30000,
print_cost=True,
is_plot=True,
lambd=0,
keep_prob=1):
'''
实现一个三层的神经网络:linear->relu->linear->relu->linear->sigmoid
:param X: 输入的数据,维度为(2,要训练/测试的数量)
:param Y: 标签,[0(蓝色)|1(红色)],维度为(1,对应的是输入的数据的标签)
:param learing_rate: 学习速率
:param num_iterations: 迭代次数
:param print_cost: 是否打印成本值
:param lambd: 正则化的超参数,实数
:param keep_prob: 随机删除节点的概率
:return:
parameters-学习后的参数
'''
grads = {}
costs = []
m = X.shape[1]
layers_dims = [X.shape[0], 20, 3, 1]
#初始化参数
parameters = reg_utils.initialize_parameters(layers_dims)
#开始学习
for i in range(0, num_iterations):
#前向传播
##是否删除随机节点
if keep_prob == 1:
###不删除随机节点
a3, cache = reg_utils.forward_propagation(X, parameters)
elif keep_prob < 1:
###随即删除节点
a3, cache = forward_propagation_with_dropout(
X, parameters, keep_prob)
else:
print('keep_prob参数错误,程序退出.')
exit
#计算成本
##是否使用二范数
if lambd == 0:
###不使用L2正则化
cost = reg_utils.compute_cost(a3, Y)
else:
###使用L2正则化
cost = compute_cost_with_regularization(a3, Y, parameters, lambd)
##反向传播
##可以同时使用L2正则化和随机删除节点,但本次实验不同时使用
assert (lambd == 0 or keep_prob == 1)
##两个参数的使用情况
if (lambd == 0 and keep_prob == 1):
###不适用L2正则化和不使用随即删除节点
grads = reg_utils.backward_propagation(X, Y, cache)
elif lambd != 0:
###使用L2正则化,不使用随即删除节点
grads = backward_propagation_with_regularization(
X, Y, cache, lambd)
elif keep_prob < 1:
###使用随即删除节点,不使用L2正则化
grads = backward_propagation_with_dropout(X, Y, cache, keep_prob)
#更新参数
parameters = reg_utils.update_parameters(parameters, grads,
learning_rate)
#记录并打印成本
if i % 1000 == 0:
##记录成本
costs.append(cost)
if (print_cost and i % 10000 == 0):
#打印成本
print('第' + str(i) + '次迭代,成本值为:' + str(cost))
#是否绘制成本曲线图
if is_plot:
plt.plot(costs)
plt.ylabel('cost')
plt.xlabel('iterations(x1,000)')
plt.title('Learning rate = ' + str(learning_rate))
plt.show()
#返回学习后的参数
return parameters
def L_layer_model(X,
Y,
layers_dims,
learning_rate=0.0075,
num_iterations=3000,
print_cost=False): #lr was 0.009
"""
Implements a L-layer neural network: [LINEAR->RELU]*(L-1)->LINEAR->SIGMOID.
Arguments:
X -- data, numpy array of shape (number of examples, num_px * num_px * 3)
Y -- true "label" vector (containing 0 if cat, 1 if non-cat), of shape (1, number of examples)
layers_dims -- list containing the input size and each layer size, of length (number of layers + 1).
learning_rate -- learning rate of the gradient descent update rule
num_iterations -- number of iterations of the optimization loop
print_cost -- if True, it prints the cost every 100 steps
Returns
parameters -- parameters learnt by the model. They can then be used to predict.
"""
np.random.seed(1)
costs = [] # keep track of cost
# Parameters initialization.
### START CODE HERE ###
parameters = initialize_parameters_deep(layers_dims)
### END CODE HERE ###
# Loop (gradient descent)
for i in range(0, num_iterations):
# Forward propagation: [LINEAR -> RELU]*(L-1) -> LINEAR -> SIGMOID.
### START CODE HERE ### (~ 1 line of code)
AL, caches = L_model_forward(X, parameters)
### END CODE HERE ###
# Compute cost.
### START CODE HERE ### (~ 1 line of code)
cost = compute_cost(AL, Y)
### END CODE HERE ###
# Backward propagation.
### START CODE HERE ### (~ 1 line of code)
grads = L_model_backward(AL, Y, caches)
### END CODE HERE ###
# Update parameters.
### START CODE HERE ### (~ 1 line of code)
parameters = update_parameters(parameters, grads, learning_rate)
### END CODE HERE ###
# Print the cost every 100 training example
if print_cost and i % 1 == 0:
print("Cost after iteration %i: %f" % (i, cost))
if print_cost and i % 100 == 0:
costs.append(cost)
# plot the cost
plt.plot(np.squeeze(costs))
plt.ylabel('cost')
plt.xlabel('iterations (per tens)')
plt.title("Learning rate =" + str(learning_rate))
plt.show()
return parameters
def model(X,Y,learning_rate = 0.3,num_iterations = 30000,print_cost = True,is_plot = True,lambd = 0,keep_prob = 1):
'''
实现一个三层的神经网络:linear->relu->linear->relu->linear->sigmoid
:param X: 输入的数据,维度为(2,要训练/测试的数量)
:param Y: 标签,[0(蓝色)|1(红色)],维度为(1,对应的是输入的数据的标签)
:param learing_rate: 学习速率
:param num_iterations: 迭代次数
:param print_cost: 是否打印成本值
:param lambd: 正则化的超参数,实数
:param keep_prob: 随机删除节点的概率
:return:
parameters-学习后的参数
'''
grads = {}
costs = []
m = X.shape[1]
layers_dims = [X.shape[0],20,3,1]
#初始化参数
parameters = reg_utils.initialize_parameters(layers_dims)
#开始学习
for i in range(0,num_iterations):
#前向传播
##是否删除随机节点
if keep_prob == 1:
###不删除随机节点
a3,cache = reg_utils.forward_propagation(X,parameters)
elif keep_prob<1:
###随即删除节点
a3 , cache = forward_propagation_with_dropout(X,parameters,keep_prob)
else:
print('keep_prob参数错误,程序退出.')
exit
#计算成本
##是否使用二范数
if lambd == 0:
###不使用L2正则化
cost = reg_utils.compute_cost(a3,Y)
else:
###使用L2正则化
cost = compute_cost_with_regularization(a3,Y,parameters,lambd)
##反向传播
##可以同时使用L2正则化和随机删除节点,但本次实验不同时使用
assert(lambd == 0 or keep_prob == 1)
##两个参数的使用情况
if(lambd == 0 and keep_prob == 1):
###不适用L2正则化和不使用随即删除节点
grads = reg_utils.backward_propagation(X,Y,cache)
elif lambd!=0:
###使用L2正则化,不使用随即删除节点
grads = backward_propagation_with_regularization(X,Y,cache,lambd)
elif keep_prob<1:
###使用随即删除节点,不使用L2正则化
grads = backward_propagation_with_dropout(X,Y,cache,keep_prob)
#更新参数
parameters = reg_utils.update_parameters(parameters,grads,learning_rate)
#记录并打印成本
if i%1000 == 0:
##记录成本
costs.append(cost)
if(print_cost and i%10000==0):
#打印成本
print('第'+str(i)+'次迭代,成本值为:'+str(cost))
#是否绘制成本曲线图
if is_plot:
plt.plot(costs)
plt.ylabel('cost')
plt.xlabel('iterations(x1,000)')
plt.title('Learning rate = '+str(learning_rate))
plt.show()
#返回学习后的参数
return parameters
def model2(X,
Y,
learning_rate=0.3,
num_iterations=30000,
print_cost=True,
is_plot=True,
lambd=0,
keep_prob=1):
grads = {}
costs = []
m = X.shape[1]
layers_dims = [X.shape[0], 20, 3, 1]
parameters = reg_utils.initialize_parameters(layers_dims)
for i in range(0, num_iterations):
# 前向传播
# 询问是否删除节点
if keep_prob == 1:
# 不随机删除节点
a3, cache = reg_utils.forward_propagation(X, parameters)
elif keep_prob < 1:
# 随即删除节点
a3, cache = forward_propagation_with_dropout(
X, parameters, keep_prob)
else:
print("keep_prob参数错误！程序退出")
exit
# 计算成本
# 是否正则化
if lambd == 0:
cost = reg_utils.compute_cost(a3, Y)
else:
cost = compute_cost_with_regularization(a3, Y, parameters, lambd)
# 本次实验不同时使用正则化和随即删除节点
assert (lambd == 0 or keep_prob == 1)
if (lambd == 0 and keep_prob == 1):
grads = reg_utils.backward_propagation(X, Y, cache)
elif lambd != 0:
grads = backward_propagation_with_regularization(
X, Y, cache, lambd)
elif keep_prob < 1:
grads = backward_propagation_with_dropout(X, Y, cache, keep_prob)
# 更新参数
parameters = reg_utils.update_parameters(parameters, grads,
learning_rate)
if i % 1000 == 0:
# 要记录的
costs.append(cost)
if (print_cost and i % 10000 == 0):
# 要输出的
print("第" + str(i) + "次迭代， 成本为：" + str(cost))
if is_plot:
plt.plot(costs)
plt.ylabel('cost')
plt.xlabel('iterations(x1,000)')
plt.title("learning rate =" + str(learning_rate))
plt.show()
return parameters
def model(X,
Y,
learning_rate=0.3,
num_iterations=30000,
print_cost=True,
lambd=0,
keep_prob=1):
grads = {}
costs = [] # to keep track of the cost
m = X.shape[1] # number of examples
layers_dims = [X.shape[0], 20, 3, 1]
# Initialize parameters dictionary.
parameters = initialize_parameters(layers_dims)
# Loop (gradient descent)
for i in range(0, num_iterations):
# Forward propagation: LINEAR -> RELU -> LINEAR -> RELU -> LINEAR -> SIGMOID.
if keep_prob == 1:
a3, cache = forward_propagation(X, parameters)
elif keep_prob < 1:
a3, cache = forward_propagation_with_dropout(
X, parameters, keep_prob)
# Cost function
if lambd == 0:
cost = compute_cost(a3, Y)
else:
cost = compute_cost_with_regularization(a3, Y, parameters, lambd)
# Backward propagation.
assert (lambd == 0 or keep_prob == 1
) # it is possible to use both L2 regularization and dropout,
# but this assignment will only explore one at a time
if lambd == 0 and keep_prob == 1:
grads = backward_propagation(X, Y, cache)
elif lambd != 0:
grads = backward_propagation_with_regularization(
X, Y, cache, lambd)
elif keep_prob < 1:
grads = backward_propagation_with_dropout(X, Y, cache, keep_prob)
# Update parameters.
parameters = update_parameters(parameters, grads, learning_rate)
# Print the loss every 10000 iterations
if print_cost and i % 10000 == 0:
print("Cost after iteration {}: {}".format(i, cost))
if print_cost and i % 1000 == 0:
costs.append(cost)
# plot the cost
plt.plot(costs)
plt.ylabel('cost')
plt.xlabel('iterations (x1,000)')
plt.title("Learning rate =" + str(learning_rate))
plt.show()
return parameters
def model(X,
Y,
learning_rate=0.3,
num_iterations=30000,
print_cost=True,
lambd=0,
keep_prob=1):
"""
implements a three layer neural network:
linear -> relu -> linear -> relu -> linear -> sigmoid
arguments:
X -- input data, shape(input size,number of examples)
Y -- true label vector, shape(output size,number of examples)
learning_rate -- learn rate of the optimization
num_iterations -- number of iterations of the optimization loop
print_cost -- if true, print the cost every 10000 iterations
lambd -- regularization hhyperparameter
keep_prob -- probability of keeping a neuron activa during drop-out
returns:
parameters -- parameters learned by the model, they can then be used to predict
"""
grads = {}
costs = []
m = X.shape[1]
layers_dims = [X.shape[0], 20, 3, 1]
#initialize parameters dictionary
parameters = initialize_parameters(layers_dims)
#loop
for i in range(0, num_iterations):
#forward
if keep_prob == 1:
a3, cache = forward_propagation(X, parameters)
elif keep_prob < 1:
a3, cache = forward_propagation_with_dropout(
X, parameters, keep_prob)
#cost function
if lambd == 0:
cost = compute_cost(a3, Y)
else:
cost = compute_cost_with_regularization(a3, Y, parameters, lambd)
#
assert (lambd == 0 or keep_prob == 1)
if lambd == 0 and keep_prob == 1:
grads = backward_propagation(X, Y, cache)
elif lambd != 0:
grads = backward_propagation_with_regularization(
X, Y, cache, lambd)
elif keep_prob < 1:
grads = backward_propagation_with_dropout(X, Y, cache, keep_prob)
#update parameters
parameters = update_parameters(parameters, grads, learning_rate)
#print the loss every 1000 iterations
if print_cost and i % 10000 == 0:
print("Cost after iteration {}: {}".format(i, cost))
if print_cost and i % 100 == 0:
costs.append(cost)
#plot the cost
plt.figure(2)
plt.plot(costs)
plt.ylabel('cost')
plt.xlabel('iterations (x1,000)')
plt.title('learning rate = ' + str(learning_rate))
plt.show()
return parameters
def model(X, Y, learning_rate=0.3, num_iterations=30000, print_cost=True, is_plot=True, lambd=0, keep_prob=1):
"""
实现一个三层的神经网络：LINEAR ->RELU -> LINEAR -> RELU -> LINEAR -> SIGMOID
参数：
X - 输入的数据，维度为(2, 要训练/测试的数量)
Y - 标签，【0(蓝色) | 1(红色)】，维度为(1，对应的是输入的数据的标签)
learning_rate - 学习速率
num_iterations - 迭代的次数
print_cost - 是否打印成本值，每迭代10000次打印一次，但是每1000次记录一个成本值
is_polt - 是否绘制梯度下降的曲线图
lambd - 正则化的超参数，实数
keep_prob - 随机删除节点的概率
返回
parameters - 学习后的参数
"""
grads = {}
costs = []
m = X.shape[1]
layers_dims = [X.shape[0], 20, 3, 1]
# 初始化参数
parameters = reg_utils.initialize_parameters(layers_dims)
# 开始学习
for i in range(0, num_iterations):
# 前向传播
##是否随机删除节点
if keep_prob == 1:
###不随机删除节点
a3, cache = reg_utils.forward_propagation(X, parameters)
elif keep_prob < 1:
###随机删除节点
a3, cache = reg_utils.forward_propagation_with_dropout(X, parameters, keep_prob)
else:
print("keep_prob参数错误！程序退出。")
exit
# 计算成本
## 是否使用二范数
if lambd == 0:
###不使用L2正则化
cost = reg_utils.compute_cost(a3, Y)
else:
###使用L2正则化
cost = reg_utils.compute_cost_with_regularization(a3, Y, parameters, lambd)
# 反向传播
##可以同时使用L2正则化和随机删除节点，但是本次实验不同时使用。
assert (lambd == 0 or keep_prob == 1)
##两个参数的使用情况
if (lambd == 0 and keep_prob == 1):
### 不使用L2正则化和不使用随机删除节点
grads = reg_utils.backward_propagation(X, Y, cache)
elif lambd != 0:
### 使用L2正则化，不使用随机删除节点
grads = reg_utils.backward_propagation_with_regularization(X, Y, cache, lambd)
elif keep_prob < 1:
### 使用随机删除节点，不使用L2正则化
grads =reg_utils. backward_propagation_with_dropout(X, Y, cache, keep_prob)
# 更新参数
parameters = reg_utils.update_parameters(parameters, grads, learning_rate)
# 记录并打印成本
if i % 1000 == 0:
## 记录成本
costs.append(cost)
if (print_cost and i % 10000 == 0):
# 打印成本
print("第" + str(i) + "次迭代，成本值为：" + str(cost))
# 是否绘制成本曲线图
if is_plot:
plt.plot(costs)
plt.ylabel('cost')
plt.xlabel('iterations (x1,000)')
plt.title("Learning rate =" + str(learning_rate))
plt.show()
# 返回学习后的参数
return parameters
def model(X,
Y,
learning_rate=0.3,
num_iterations=30000,
print_cost=True,
lambd=0,
keep_prob=1):
"""
实现一个三层神经网络模型: LINEAR->RELU->LINEAR->RELU->LINEAR->SIGMOID.
可以：-- 不使用正则化
-- 使用 L2 正则化
-- 使用 dropout 正则化
:param X: 输入数据, of shape (input size, number of examples)
:param Y: 正确的标签向量 (1 for blue dot / 0 for red dot), of shape (output size, number of examples)
:param learning_rate: 优化的学习率
:param num_iterations: 优化循环的迭代次数
:param print_cost: If True, print the cost every 10000 iterations
:param lambd: 正则化的超参数 λ
:param keep_prob: drop-out 正则化中随机保留节点的概率
Returns:
parameters -- 神经网络模型学习得到的参数，可以用来预测
"""
grads = {}
costs = [] # to keep track of the cost
m = X.shape[1] # number of examples
layers_dims = [X.shape[0], 20, 3, 1]
# 初始化参数字典
parameters = initialize_parameters(layers_dims)
# 学习循环（梯度下降）
for i in range(0, num_iterations):
# 前向传播: LINEAR -> RELU -> LINEAR -> RELU -> LINEAR -> SIGMOID.
# 不使用 dropout 正则化
if keep_prob == 1:
a3, cache = forward_propagation(X, parameters)
# 使用 dropout 正则化
elif keep_prob < 1:
a3, cache = forward_propagation_with_dropout(
X, parameters, keep_prob)
# 计算 cost
# 不使用 L2 正则化
if lambd == 0:
cost = compute_cost(a3, Y)
# 使用 L2 正则化
else:
cost = compute_cost_with_regularization(a3, Y, parameters, lambd)
# 后向传播
assert (lambd == 0 or keep_prob == 1
) # 可以同时使用 L2 正则化和 dropout, 但是本程序只使用其中一个或不使用正则化
# 不使用正则化
if lambd == 0 and keep_prob == 1:
grads = backward_propagation(X, Y, cache)
# 使用 L2 正则化
elif lambd != 0:
grads = backward_propagation_with_regularization(
X, Y, cache, lambd)
# 使用 dropout 正则化
elif keep_prob < 1:
grads = backward_propagation_with_dropout(X, Y, cache, keep_prob)
# 更新参数
parameters = update_parameters(parameters, grads, learning_rate)
# Print the loss every 10000 iterations
if print_cost and i % 10000 == 0:
print("Cost after iteration {}: {}".format(i, cost))
if print_cost and i % 1000 == 0:
costs.append(cost)
# 绘制 cost 变化曲线图
plt.plot(costs)
plt.ylabel('cost')
plt.xlabel('iterations (x1,000)')
plt.title("Learning rate =" + str(learning_rate))
plt.show()
return parameters
def model(X,
Y,
learning_rate=0.3,
num_iterations=30000,
print_cost=True,
lambd=0,
keep_prob=1):
"""
Implements a three-layer neural network: LINEAR->RELU->LINEAR->RELU->LINEAR->SIGMOID.
Arguments:
X -- input data, of shape (input size, number of examples)
Y -- true "label" vector (1 for blue dot / 0 for red dot), of shape (output size, number of examples)
learning_rate -- learning rate of the optimization
num_iterations -- number of iterations of the optimization loop
print_cost -- If True, print the cost every 10000 iterations
lambd -- regularization hyperparameter, scalar
keep_prob - probability of keeping a neuron active during drop-out, scalar.
Returns:
parameters -- parameters learned by the model. They can then be used to predict.
"""
grads = {}
costs = [] # to keep track of the cost
m = X.shape[1] # number of examples
layers_dims = [X.shape[0], 20, 3, 1]
# Initialize parameters dictionary.
parameters = initialize_parameters(layers_dims)
# Loop (gradient descent)
for i in range(0, num_iterations):
# Forward propagation: LINEAR -> RELU -> LINEAR -> RELU -> LINEAR -> SIGMOID.
if keep_prob == 1:
a3, cache = forward_propagation(X, parameters)
elif keep_prob < 1:
a3, cache = forward_propagation_with_dropout(
X, parameters, keep_prob)
# Cost function
if lambd == 0:
cost = compute_cost(a3, Y)
else:
cost = compute_cost_with_regularization(a3, Y, parameters, lambd)
# Backward propagation.
assert (lambd == 0 or keep_prob == 1
) # it is possible to use both L2 regularization and dropout,
# but this assignment will only explore one at a time
if lambd == 0 and keep_prob == 1:
grads = backward_propagation(X, Y, cache)
elif lambd != 0:
grads = backward_propagation_with_regularization(
X, Y, cache, lambd)
elif keep_prob < 1:
grads = backward_propagation_with_dropout(X, Y, cache, keep_prob)
# Update parameters.
parameters = update_parameters(parameters, grads, learning_rate)
# Print the loss every 10000 iterations
if print_cost and i % 10000 == 0:
print("Cost after iteration {}: {}".format(i, cost))
if print_cost and i % 1000 == 0:
costs.append(cost)
# plot the cost
plt.plot(costs)
plt.ylabel('cost')
plt.xlabel('iterations (x1,000)')
plt.title("Learning rate =" + str(learning_rate))
plt.show()
return parameters
# Initialize parameters dictionary.
parameters = initialize_parameters(layers_dims)
# Loop (gradient descent)
for i in range(0, num_iterations):
# Forward propagation: LINEAR -> RELU -> LINEAR -> RELU -> LINEAR -> SIGMOID.
if keep_prob == 1:
a3, cache = forward_propagation(X, parameters)
elif keep_prob < 1:
a3, cache = forward_propagation_with_dropout(X, parameters, keep_prob)
# Cost function
if lambd == 0:
cost = compute_cost(a3, Y)
else:
cost = compute_cost_with_regularization(a3, Y, parameters, lambd)
# Backward propagation.
assert(lambd==0 or keep_prob==1) # it is possible to use both L2 regularization and dropout,
# but this assignment will only explore one at a time
if lambd == 0 and keep_prob == 1:
grads = backward_propagation(X, Y, cache)
elif lambd != 0:
grads = backward_propagation_with_regularization(X, Y, cache, lambd)
elif keep_prob < 1:
grads = backward_propagation_with_dropout(X, Y, cache, keep_prob)
# Update parameters.
parameters = update_parameters(parameters, grads, learning_rate)