写一个代码来计算向后传播，深度学习课程由Andrew NG

Question

所以我参加了Andrew NG在coursera上的深度学习AI课程。我目前正在完成第2周的最后一项任务。我到达了必须编写正向和反向传播函数的部分。我设法编写了fwd_propagation函数，它相当简单。这是下面的代码：

def fwd_propagate(w,b,X,y):
m = X.shape[1]
A = sigmoid(np.dot(w.T,X)+b)
J = (-1/m)*np.sum(y * np.log(A) + (1-y) * np.log(1-A))
return J

现在我必须编写bwd_propagation函数，但我不知道从哪里以及如何开始。有人能帮我解释一下我应该写什么吗。

这是我迄今为止写的关于测试的所有内容。

import numpy as np
import matplotlib.pyplot as plt
import h5py
import scipy
from PIL import Image
from scipy import ndimage

%matplotlib inline
def load_dataset():
train_dataset = h5py.File('C:/Users/Univ/Desktop/ML Intern/Logistic-Regression-with-a-Neural-Network-mindset-master/train_catvnoncat.h5', "r")
train_set_x_orig = np.array(train_dataset["train_set_x"][:]) # your train set features
train_set_y_orig = np.array(train_dataset["train_set_y"][:]) # your train set labels
test_dataset = h5py.File('C:/Users/Univ/Desktop/ML Intern/Logistic-Regression-with-a-Neural-Network-mindset-master/test_catvnoncat.h5', "r")
test_set_x_orig = np.array(test_dataset["test_set_x"][:]) # your test set features
test_set_y_orig = np.array(test_dataset["test_set_y"][:]) # your test set labels
classes = np.array(test_dataset["list_classes"][:]) # the list of classes

train_set_y_orig = train_set_y_orig.reshape((1, train_set_y_orig.shape[0]))
test_set_y_orig = test_set_y_orig.reshape((1, test_set_y_orig.shape[0]))

return train_set_x_orig, train_set_y_orig, test_set_x_orig, test_set_y_orig, classes
train_set_x_orig, train_set_y, test_set_x_orig, test_set_y, classes = load_dataset()
index = 25
plt.imshow(train_set_x_orig[index])
print ("y = " + str(train_set_y[:,index]) + ", it's a '" + classes[np.squeeze(train_set_y[:,index])].decode("utf-8") +  "' picture.")
print(str(train_set_y.shape[1]) + " This is the amount of elements in the training set")
print(str(test_set_y.shape[1]) + " This is the amount of elements in the test set")
print(str(train_set_x_orig.shape[1]) + " This is the Num_px")
print(f"{train_set_x_orig.shape[1]} This is the Num_px")
print(train_set_x_orig.shape)
X_flatten1 = train_set_x_orig.reshape(train_set_x_orig.shape[0], -1).T
X_flatten2 = train_set_y.reshape(train_set_y.shape[0], -1).T 
X_flatten3 = test_set_x_orig.reshape(test_set_x_orig.shape[0], -1).T 
X_flatten4 = test_set_y.reshape(test_set_y.shape[0], -1).T 
print(X_flatten1)
print(X_flatten2)
print(X_flatten3)
print(X_flatten4)
print(X_flatten1.shape)
print(X_flatten2.shape)
print(X_flatten3.shape)
print(X_flatten4.shape)
print(" Let's standardize our date")
train_set_x = X_flatten1/256
test_set_x = X_flatten3/256
print(train_set_x)
print(test_set_x)
def sigmoid(x):
s = 1/(1+np.exp(-x))
return s
print ("sigmoid(0) = " + str(sigmoid(0)))
print ("sigmoid(9.2) = " + str(sigmoid(9.2)))
def initialize_with_zeros(dim):
shp=(dim,1)
w = np.zeros(shp)
b = 0
assert(w.shape == (dim, 1))
assert(isinstance(b, float) or isinstance(b, int))
return w,b
dim = 2
w, b = initialize_with_zeros(dim)
print ("w = " + str(w))
print ("b = " + str(b))
def fwd_propagate(w,b,X,y):
m = X.shape[1]
A = sigmoid(np.dot(w.T,X)+b)
J = (-1/m)*np.sum(y * np.log(A) + (1-y) * np.log(1-A))
return J

Answer 1

下一步是计算反向传播的导数：

dw = 1/m*(np.dot(X, ((A-Y).T)))
db = 1/m*(np.sum(A-Y))