1.1 欢迎

1. Neural Networks and Deep Learning.
2. Improving Deep Neural Networks: Hyper-parameter tuning, Regularization and Optimization.
3. Structuring your Machine Learning project.
4. Convolutional Neural Networks.
5. Neural Language Processing: Building sequence models.

1.2 什么是神经网络

1.3 用神经网络进行监督学习 【supervised learning】

1.4 为什么深度学习会兴起？

Data & Computation & Algorithms

1.5 关于这门课

• Introduction
• Basics of Neural Network Programming
• One hidden layer Neural Networks
• Deep Neural Networks

1.6 课程资源

2.1 二分分类

2.2 Logistic 回归【Logistic Regression】

$\hat{y}=\sigma (w^{T}x+b)$ using sigmoid function $\sigma =\frac{1}{1+e^{-z}}$.

2.3 logistic 回归损失函数

$\mathcal{L}(\hat{y},y)=\frac{1}{2}(\hat{y}-y{)}^2$ ×
$\mathcal{L}(\hat{y},y)=-(y\log \hat{y}+(1-y)\log (1-\hat{y}))$ √

2.4 梯度下降法

$J(w,b)=\frac{1}{m}{\sum }_{i=1}^m\mathcal{L}({\hat{y}}^{(i)},y^{(i)})=-\frac{1}{m}{\sum }_{i=1}^m(y^{(i)}\log {\hat{y}}^{(i)}+(1-y^{(i)})\log (1-{\hat{y}}^{(i)}))$

$w:=w=\alpha \frac{dJ}{dw}$ in code, $dw$ means $\frac{dJ}{dw}$.

2.5 导数

2.6 更多导数的例子

2.7 计算图

2.8 使用计算图求导

2.9 logistic 回归中的梯度下降法

2.10 m 个样本的梯度下降

2.11 向量化

import numpy as np
a = np.array([1, 2, 3, 4])

import time

a = np.random.rand(10000000)
b = np.random.rand(10000000)

tic = time.time()
c = np.dot(a, b)
toc = time.time()

print("vectorized version:"+str(100*(toc-tic))+"ms")

c = 0
tic = time.time()
for i in range(10000000):
  c += a[i] * b[i]
toc = time.time()

print("for loop version:"+str(100*(toc-tic))+"ms")

vectorized version: 1.7638683319091797ms
for loop version: 431.3936233520508ms

参考文献：

https://www.coursera.org/deeplearning-ai