Coursera-DeepLearning-Part4-Week2-take away

Course4- Convolutional Neural Networks (CNN)

学习目标

• Understand multiple foundational papers of convolutional neural networks
• Analyze the dimensionality reduction of a volume in a very deep network
• Understand and Implement a Residual network
• Build a deep neural network using Keras
• Implement a skip-connection in your network
• Clone a repository from github and use transfer learning

---

一. Case Studies

1. Why look at case studies?

---

2. Classic Networks

---

3. ResNets

--

4. Why ResNets Work

---

5. Networks in Networks and 1*1 Convolutions

---

6. Inception Network Motivation

---

7. Inception Network

---

二. Practical advices for using ConvNets

1. Using Open-Source Implementation

建议使用GitHub中的open source资料

---

2. Transfer Learning

当数据量很小的时候，NG会选择用transfer learning, 如果自身的数据量很大，计算能力很强时，当然也可以从头开始初始化参数。

然而，COMPUTER VISION 通常总是需要借助Transfer learning已经调好的参数，在最后几层，换成自己的全连接层和输出层。

---

3. Data Augmentation

---

4. State of Computer

---

三. Quiz

---

四. Programming Assignment

1. Program-1 Keras Tutorial- The happy house

---

2. Program-2 Residual Networks

---

读论文的顺序：
AlexNet -> VGG-16 -> LeNet
从简单到复杂（阅读难易程度）