Basic Generative Adversarial Networks
0 likes853 views
The document discusses Generative Adversarial Networks (GANs), a type of generative model proposed by Ian Goodfellow in 2014. GANs use two neural networks, a generator and discriminator, that compete against each other. The generator produces synthetic data to fool the discriminator, while the discriminator learns to distinguish real from synthetic data. GANs have been used successfully to generate realistic images when trained on large datasets. Examples mentioned include Pix2Pix for image-to-image translation and STACKGAN for text-to-image generation.
![Generative Model 1D example
[0, 1]
http://m.blog.naver.com/atelierjpro/22098475
8512](https://image.slidesharecdn.com/gan-170613071027/85/Basic-Generative-Adversarial-Networks-3-320.jpg)
![Generative Adversarial Nets
GAN started by Ian Goodfellow
[https://arxiv.org/abs/1406.2661]
GAN used for generating realistic data
( usually, for images )](https://image.slidesharecdn.com/gan-170613071027/85/Basic-Generative-Adversarial-Networks-4-320.jpg)
Basic Generative Adversarial Networks
- 1. GAN Generative Adversarial Networks from scratch For 수알못(Na)
- 3. Generative Model 1D example [0, 1] http://m.blog.naver.com/atelierjpro/22098475 8512
- 4. Generative Adversarial Nets GAN started by Ian Goodfellow [https://arxiv.org/abs/1406.2661] GAN used for generating realistic data ( usually, for images )
- 5. Generative Adversarial Nets By Ian Goodfellow ( 2014. 06 ) OpenAI https://arxiv.org/pdf/1406.2661.pdf
- 6. 경찰과 도둑 (유즈맵 아님)
- 7. GAN : How does it works? Random NoiseInput : Random Noise Output: Realistic Image Discriminator Network
- 8. GAN : How does it works? < Traditional Training Model >
- 9. GAN : How does it works? < GAN Training Model > MAIN IDEA : Just showing lot’s of images Initial Trained
- 10. GAN : How does it works? < GAN Training Model > MAIN IDEA : Just showing lot’s of images
- 11. Generative Model 1D example http://m.blog.naver.com/atelierjpro/22098475 8512
- 12. Generative Model 1D example http://m.blog.naver.com/atelierjpro/22098475 8512 g = G(z) # z is uniform distribution Generative Function Input : Uniform distribution Output : Real Data
- 13. Generative Model 1D example http://m.blog.naver.com/atelierjpro/22098475 8512 g = G(z) # z is uniform distribution Generative Function : G is neural network Input : Uniform distribution Output : Real Data Not Trained Model (yellow)
- 14. Generative Model 1D example http://m.blog.naver.com/atelierjpro/22098475 8512 (0 || 1) = D(d) # d is data ( from gen func or real data ) Discriminator : D is neural network Input : Data Output : real or fake Black is D ( input from not trained G ) Yellow is G ( not trained ) - 1 Red is input of G Blue is Real Data Discriminator가 잘 구분함 ( G와 Real이 완전이 분리)
- 15. Generative Model 1D example http://m.blog.naver.com/atelierjpro/22098475 8512 (0 || 1) = D(d) # d is data ( from gen func or real data ) Discriminator : D is neural network Input : Data Output : real or fake Black is D ( input from not trained G ) Yellow is G ( not trained ) - 2 Red is input of G Blue is Real Data Discriminator가 잘 구분 못함 ( G와 Real이 섞여있음)
- 16. GAN 1D test
- 17. GAN EXAMPLES STACKGAN, Pix2Pix etc..
- 18. STACKGAN MODEL Text to Image ( not searching from db, drawing from probability distribution )
- 20. CycleGAN https://github.com/junyanz/CycleGAN Image to Image (Zebra는 다른 환경에서 살기 때문에 배경도 Africa처럼 보이는 것을 확인가능)
- 21. Apple’s First Paper Learning from Simulated and Unsupervised Images through Adversarial Training https://arxiv.org/abs/1612.07828 Image to Image Fake to REAL
- 22. GAME THEORY 한장 요약 존 폰 노이만 (사기 캐릭)
- 23. GAN Architecture
- 24. GAN Architecture AdamOptimizer(g_loss) AdamOptimizer(d_loss) loss = sigmoid_cross_entropy(D_out, labels) # D_out = ( 0 || 1) p = sigmoid(logits) loss =cross_entropy(logits, labels * 0.9) # training skill d_loss = cross_entropy(logits, labels) g_loss = cross_entropy(logits, flipped_labels Flipped Labels: 0 -> 1 | 1 -> 0
- 25. 참고보면 좋은 자료 https://www.slideshare.net/ssuser7e10e4/wasserstein-gan-i : 전 수학을 몰라욧! ( GAN 수학 ) Udacity deep learning nanodegree https://en.wikipedia.org/wiki/Generative_model ( Thanks WIKI )