本次作业相对简单,稍微比较复杂的是grad_check的部分,代码贴上来如下:
def gradient_check_n(parameters, gradients, X, Y, epsilon=1e-7):
"""
Checks if backward_propagation_n computes correctly the gradient of the cost output by forward_propagation_n
Arguments:
parameters -- python dictionary containing your parameters "W1", "b1", "W2", "b2", "W3", "b3":
grad -- output of backward_propagation_n, contains gradients of the cost with respect to the parameters.
x -- input datapoint, of shape (input size, 1)
y -- true "label"
epsilon -- tiny shift to the input to compute approximated gradient with formula(1)
Returns:
difference -- difference (2) between the approximated gradient and the backward propagation gradient
"""
# Set-up variables
parameters_values, _ = dictionary_to_vector(parameters)
grad = gradients_to_vector(gradients)
num_parameters = parameters_values.shape[0]
J_plus = np.zeros((num_parameters, 1))
J_minus = np.zeros((num_parameters, 1))
gradapprox = np.zeros((num_parameters, 1))
# Compute gradapprox
for i in range(num_parameters):
# Compute J_plus[i]. Inputs: "parameters_values, epsilon". Output = "J_plus[i]".
# "_" is used because the function you have to outputs two parameters but we only care about the first one
### START CODE HERE ### (approx. 3 lines)
new_theta_plus = np.copy(parameters_values)
new_theta_plus[i, 0] = new_theta_plus[i, 0] + epsilon
J_plus[i, 0], _ = forward_propagation_n(X, Y, vector_to_dictionary(new_theta_plus))
### END CODE HERE ###
# Compute J_minus[i]. Inputs: "parameters_values, epsilon". Output = "J_minus[i]".
### START CODE HERE ### (approx. 3 lines)
new_theta_minus = np.copy(parameters_values)
new_theta_minus[i, 0] = new_theta_minus[i, 0] - epsilon
J_minus[i, 0], _ = forward_propagation_n(X, Y, vector_to_dictionary(new_theta_minus))
### END CODE HERE ###
# Compute gradapprox[i]
### START CODE HERE ### (approx. 1 line)
gradapprox[i, 0] = (J_plus[i, 0] - J_minus[i, 0]) / (2 * epsilon)
### END CODE HERE ###
# Compare gradapprox to backward propagation gradients by computing difference.
### START CODE HERE ### (approx. 1 line)
# Step 1'
denominator = np.linalg.norm(grad) + np.linalg.norm(gradapprox)
# Step 2'
numerator = np.linalg.norm(grad - gradapprox)
# Step 3'
difference = numerator / denominator
### END CODE HERE ###
if difference > 1.2e-7:
print(
"\033[93m" + "There is a mistake in the backward propagation! difference = " + str(difference) + "\033[0m")
else:
print(
"\033[92m" + "Your backward propagation works perfectly fine! difference = " + str(difference) + "\033[0m")
return difference
运行后会发现产生了0.28的difference,修改了back_propagation中故意设置的两个错误后,偏差如下:
1.1890417878779317e-07
最开始我改了back_propagation中的两个错误后,发现仍提示difference过大,经上网查证,发现这个difference是对的,只是大家都把difference的阈值给做了调整。。。所以我也把阈值改成了1.2e-7(作业提供的代码中阈值为1e-7)。但这个difference和epsilon已经是一个量级了,偏差不算小,但应该也在可接受的范围内。
另外一个很迷惑的问题是,相同的代码在我的电脑上得到的偏差为1.189blabla,别人得到的就是1.188blabla,我在pycharm上和jupyter notebook上都进行了验证,发现和编译器好像没有关系。。难道这玩意和CPU有关吗。