李宏毅GAN学习笔记（01）

GAN Lecture 1

---

Yann LeCun: Adversarial training is the coolest thing since sliced bread.

Outline

• Basic Idea of GAN
• GAN as Structured Learning
• Can Generator Learn by Itself?
• Can Discriminator Generate?
• A Little Bit Theory

Basic Idea of GAN

Generation

Generation: it is a neural network or a function.

input vector: each dimension of input vector represents some characterstics.

Discriminator

Discriminator: it is a neural network or a function.

output scalar: large value means real, smaller value means fake.

GAN Algorithm

---

• Initialize generator and discriminator

• In each training iteration：

  Step 1: fix generator G, and update discriminator D
  

  Step 2: fix discriminator D, and update generator G
  

---

Initialize ${\theta }_d$ for D and ${\theta }_g$ for G

• In each training iteration:

  • Sample m examples { x 1 , x 2 , … , x m } \{ x^1, x^2, \dots, x^m \} {x1,x2,…,xm} from database
  • Sample m noise samples { z 1 , z 2 , … , z m } \{ z^1, z^2, \dots, z^m \} {z1,z2,…,zm} from a distrubution
  • Obtaining generated data { x ~ 1 , x ~ 2 , … , x ~ m } ,   x ~ i = G ( z i ) \{ \widetilde{x}^1, \widetilde{x}^2, \dots, \widetilde{x}^m \}, \, \widetilde{x}^i=G(z^i) {x 1,x 2,…,x m},x i=G(zi)
  • Update discriminator parameters ${\theta }_d$ to maximize
    • $\tilde{V}=\frac{1}{m}{\sum }_{i=1}^{m}logD(x^i)+\frac{1}{m}{\sum }_{i=1}^{m}log(1-D({\tilde{x}}^i))$
    • ${\theta }_d\leftarrow {\theta }_d+\eta {▽}_{\tilde{V}({\theta }_d)}$

  Learning D

  • Sample m noise samples { z 1 , z 2 , … , z m } \{z^1, z^2, \dots, z^m \} {z1,z2,…,zm} from a distribution
  • Update generator parameters ${\theta }_g$ to maxmize
    • $\tilde{V}=\frac{1}{m}{\sum }_{i=1}^{m}log(D(G(z^i)))$
    • ${\theta }_g\leftarrow {\theta }_g+\eta {▽}_{\tilde{V}({\theta }_g)}$

  Learning G

GAN as Structured Learning

Structured Learning Approach

• Bottom Up: Learn to generate the object at the component level

  缺乏大局观

• Top Down: Evaluating the whole object, and find the best one

$BottomUp+TopDown\Rightarrow GAN$

Can Generator Learn by Itself?

若使用Auto-Encoder技术作为Generator，那么需要更深的网络结构。

Can Discriminator Generate?

Discriminator – Training

• General Algorithm
  • Given a set of positive example, randomly generate a set of negative examples.
• In each iteration
  • Learn a discriminator D that can discriminate positive and negative examples.
  • Generate negative examples by discriminator D: $\tilde{x}=argma{x}_{x\in X}D(x)$

Generator v.s. Discriminator

• Generator
  • Pros:
    • Easy to generate even with deep model.
  • Cons:
    • Imitate the appearance.
    • Hard to learn the correlation between components.

---

• Discriminator
  • Pros:
    • Considering the big picture.
  • Cons:
    • Generation is not always feasible, especially when your model is deep.
    • How to do negative sampling?

Discriminator + Generator – Training

• General Algorithm
  • Given a set of positive example, randomly generate a set of negative examples.
• In each iteration
  • Learn a discriminator D that can discriminate positive and negative examples.
  • Generate negative examples by discriminator D: $\tilde{x}=argma{x}_{x\in X}D(x)\iff G\to \tilde{x}$