SlideShare a Scribd company logo
Image Generation with Tensorflow
Cenk Bircano˘glu
Boyner Group/Bah¸ce¸sehir Uni
cenk.bircanoglu@gmail.com
April 21, 2018
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 1 / 44
Overview
1 Convolutional Neural Networks
2 What are Generative Adverserial Networks?
3 Common GAN Architectures and Applications
4 Implementing GANs with Tensorflow
5 Question&Answer
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 2 / 44
Convolutional Neural Networks
Source: https://www.mathworks.com/discovery/convolutional-neural-network.html
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 3 / 44
Deceiving the CNN
Source: https://blog.openai.com/adversarial-example-research/
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 4 / 44
What are Generative Adverserial Networks?
Source: https://deeplearning4j.org/generative-adversarial-network
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 5 / 44
GAN Training
Source: Goodfellow, 20017
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 6 / 44
Generator
Source: http://arxiv.org/abs/1511.06434
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 7 / 44
Disriminator
Source: http://arxiv.org/abs/1511.06434
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 8 / 44
Common GAN Applications
Generating High-Quality
Images
Semi-supervised learning
Ensembles
Clustering
Image blending
Image Inpainting
Joint Probability
Super-Resolution
De-occlusion
Improving Classification &
Recognition
Semantic Segmentation
Object Detection
RNN-GANs
Conditional Adversarial Nets
Video Prediction & Generation
Texture Synthesis & Style
Transfer
Image Translation
GAN Theory
3-Dimensional GANs
Music Face Generation &
Editing
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 9 / 44
GAN Results
Source: https://kaonashi-tyc.github.io/2017/04/06/zi2zi.html, https://github.com/hanzhanggit/StackGAN,
https://arxiv.org/pdf/1708.05509.pdf, https://arxiv.org/pdf/1705.09368.pdf, https://github.com/david-gpu/srez,
http://research.nvidia.com/sites/default/files/pubs/2017-10 Progressive-Growing-of/karras2018iclr-paper.pdf,
https://github.com/Guim3/IcGAN
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 10 / 44
Generative Adversarial Networks - 2014
The first paper
Two-player game that the generator
is trained to generate images from
inputed noises to fool the
discriminator while the discriminator
is trained to well discriminate real
samples and fake samples
E real(log(D)) + E fake(log(1-D))
https://papers.nips.cc/paper/5423-generative-adversarial-nets.pdf
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 11 / 44
Conditional Generative Adversarial Nets - 2014
Source: http://guimperarnau.com/blog/2017/03/Fantastic-GANs-and-where-to-find-them
https://arxiv.org/pdf/1511.06434.pdf
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 12 / 44
cGAN Results
Source: https://arxiv.org/pdf/1411.1784.pdf
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 13 / 44
Unsupervised Representation Learning with Deep
Convolutional Generative Adversarial Networks - 2015
Source: https://arxiv.org/pdf/1511.06434.pdf
https://arxiv.org/pdf/1511.06434.pdfCenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 14 / 44
StackGAN: Text to Photo-realistic Image Synthesis with
Stacked Generative Adversarial Networks - 2016
Source: https://arxiv.org/pdf/1612.03242.pdf
https://arxiv.org/pdf/1612.03242.pdf
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 15 / 44
StackGAN Results
Source: https://arxiv.org/pdf/1612.03242.pdf
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 16 / 44
InfoGAN: Interpretable Representation Learning by
Information Maximizing Generative Adversarial Nets - 2016
Source: https://towardsdatascience.com/infogan-generative-adversarial-networks-part-iii-380c0c6712cd
https://arxiv.org/pdf/1606.03657.pdf
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 17 / 44
InfoGAN Results
Source: https://arxiv.org/pdf/1606.03657.pdf
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 18 / 44
Wasserstein GAN - 2017
Source: https://arxiv.org/pdf/1701.07875.pdf
https://arxiv.org/pdf/1701.07875.pdf
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 19 / 44
Improved Training of Wasserstein GANs - 2017
Source: https://arxiv.org/pdf/1704.00028.pdf
https://arxiv.org/pdf/1704.00028.pdf
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 20 / 44
WGAN-GP Results - 2017
Source: https://arxiv.org/pdf/1704.00028.pdf
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 21 / 44
BEGAN: Boundary Equilibrium Generative Adversarial
Networks - 2017
Source: https://arxiv.org/pdf/1703.10717.pdf
https://arxiv.org/pdf/1703.10717.pdf
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 22 / 44
BEGAN Results
Source: https://arxiv.org/pdf/1703.10717.pdf
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 23 / 44
Progressive Growing of GANs for Improved Quality,
Stability, and Variation - 2017
Source: https://arxiv.org/pdf/1710.10196.pdf
https://arxiv.org/pdf/1710.10196.pdf
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 24 / 44
ProGAN Results
Source: https://arxiv.org/pdf/1710.10196.pdf
https://arxiv.org/pdf/1710.10196.pdf
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 25 / 44
Unpaired Image-to-Image Translation using
Cycle-Consistent Adversarial Networks - 2017
Source: https://arxiv.org/pdf/1703.10593.pdf
https://arxiv.org/pdf/1703.10593.pdf
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 26 / 44
CycleGAN Results
Source: https://arxiv.org/pdf/1703.10593.pdf
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 27 / 44
Implementing GANs with Tensorflow
Generator Network
Discriminator Network
Initialize Graph
Loss, Optimizer, Weights
Training
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 28 / 44
Discriminator in Tensorflow
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 29 / 44
First Convolutional Layer in Discriminator
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 30 / 44
Second Convolutional Layer in Discriminator
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 31 / 44
First Fully Connected Layer in Discriminator
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 32 / 44
Second Fully Connected Layer in Discriminator
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 33 / 44
Generator in Tensorflow
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 34 / 44
First Convolutional Layer in Generator
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 35 / 44
Second Convolutional Layer in Generator
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 36 / 44
Third Convolutional Layer in Generator
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 37 / 44
Fourth Convolutional Layer in Generator
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 38 / 44
Initialize Graph
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 39 / 44
Loss, Optimizer, Weights in Tensorflow
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 40 / 44
Training
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 41 / 44
Results
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 42 / 44
GAN Training Tips & Tricks
Normalize the inputs
A modified loss function
Use a spherical Z
BatchNorm
Avoid Sparse Gradients:
ReLU, MaxPool
Use Soft and Noisy Labels
DCGAN / Hybrid Models
Use stability tricks from RL
Use the ADAM Optimizer
Track failures early
Dont balance loss via statistics
If you have labels, use them
Add noise to inputs, decay over
time
Train discriminator more
Batch Discrimination
Discrete variables in Conditional
GANs
Use Dropouts in G in both train
and test phase
https://github.com/soumith/ganhacks
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 43 / 44
Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 44 / 44

More Related Content

Recently uploaded

Professional Data Engineer Certification Exam Guide  _  Learn  _  Google Clou...
Professional Data Engineer Certification Exam Guide  _  Learn  _  Google Clou...Professional Data Engineer Certification Exam Guide  _  Learn  _  Google Clou...
Professional Data Engineer Certification Exam Guide  _  Learn  _  Google Clou...Domenico Conte
 
Jpolillo Amazon PPC - Bid Optimization Sample
Jpolillo Amazon PPC - Bid Optimization SampleJpolillo Amazon PPC - Bid Optimization Sample
Jpolillo Amazon PPC - Bid Optimization SampleJames Polillo
 
Computer Presentation.pptx ecommerce advantage s
Computer Presentation.pptx ecommerce advantage sComputer Presentation.pptx ecommerce advantage s
Computer Presentation.pptx ecommerce advantage sMAQIB18
 
一比一原版(CBU毕业证)不列颠海角大学毕业证成绩单
一比一原版(CBU毕业证)不列颠海角大学毕业证成绩单一比一原版(CBU毕业证)不列颠海角大学毕业证成绩单
一比一原版(CBU毕业证)不列颠海角大学毕业证成绩单nscud
 
一比一原版(ArtEZ毕业证)ArtEZ艺术学院毕业证成绩单
一比一原版(ArtEZ毕业证)ArtEZ艺术学院毕业证成绩单一比一原版(ArtEZ毕业证)ArtEZ艺术学院毕业证成绩单
一比一原版(ArtEZ毕业证)ArtEZ艺术学院毕业证成绩单vcaxypu
 
一比一原版(RUG毕业证)格罗宁根大学毕业证成绩单
一比一原版(RUG毕业证)格罗宁根大学毕业证成绩单一比一原版(RUG毕业证)格罗宁根大学毕业证成绩单
一比一原版(RUG毕业证)格罗宁根大学毕业证成绩单vcaxypu
 
一比一原版(UPenn毕业证)宾夕法尼亚大学毕业证成绩单
一比一原版(UPenn毕业证)宾夕法尼亚大学毕业证成绩单一比一原版(UPenn毕业证)宾夕法尼亚大学毕业证成绩单
一比一原版(UPenn毕业证)宾夕法尼亚大学毕业证成绩单ewymefz
 
tapal brand analysis PPT slide for comptetive data
tapal brand analysis PPT slide for comptetive datatapal brand analysis PPT slide for comptetive data
tapal brand analysis PPT slide for comptetive datatheahmadsaood
 
Algorithmic optimizations for Dynamic Levelwise PageRank (from STICD) : SHORT...
Algorithmic optimizations for Dynamic Levelwise PageRank (from STICD) : SHORT...Algorithmic optimizations for Dynamic Levelwise PageRank (from STICD) : SHORT...
Algorithmic optimizations for Dynamic Levelwise PageRank (from STICD) : SHORT...Subhajit Sahu
 
Introduction-to-Cybersecurit57hhfcbbcxxx
Introduction-to-Cybersecurit57hhfcbbcxxxIntroduction-to-Cybersecurit57hhfcbbcxxx
Introduction-to-Cybersecurit57hhfcbbcxxxzahraomer517
 
Using PDB Relocation to Move a Single PDB to Another Existing CDB
Using PDB Relocation to Move a Single PDB to Another Existing CDBUsing PDB Relocation to Move a Single PDB to Another Existing CDB
Using PDB Relocation to Move a Single PDB to Another Existing CDBAlireza Kamrani
 
Investigate & Recover / StarCompliance.io / Crypto_Crimes
Investigate & Recover / StarCompliance.io / Crypto_CrimesInvestigate & Recover / StarCompliance.io / Crypto_Crimes
Investigate & Recover / StarCompliance.io / Crypto_CrimesStarCompliance.io
 
哪里卖(usq毕业证书)南昆士兰大学毕业证研究生文凭证书托福证书原版一模一样
哪里卖(usq毕业证书)南昆士兰大学毕业证研究生文凭证书托福证书原版一模一样哪里卖(usq毕业证书)南昆士兰大学毕业证研究生文凭证书托福证书原版一模一样
哪里卖(usq毕业证书)南昆士兰大学毕业证研究生文凭证书托福证书原版一模一样axoqas
 
Tabula.io Cheatsheet: automate your data workflows
Tabula.io Cheatsheet: automate your data workflowsTabula.io Cheatsheet: automate your data workflows
Tabula.io Cheatsheet: automate your data workflowsalex933524
 
一比一原版(CBU毕业证)卡普顿大学毕业证成绩单
一比一原版(CBU毕业证)卡普顿大学毕业证成绩单一比一原版(CBU毕业证)卡普顿大学毕业证成绩单
一比一原版(CBU毕业证)卡普顿大学毕业证成绩单nscud
 
做(mqu毕业证书)麦考瑞大学毕业证硕士文凭证书学费发票原版一模一样
做(mqu毕业证书)麦考瑞大学毕业证硕士文凭证书学费发票原版一模一样做(mqu毕业证书)麦考瑞大学毕业证硕士文凭证书学费发票原版一模一样
做(mqu毕业证书)麦考瑞大学毕业证硕士文凭证书学费发票原版一模一样axoqas
 
一比一原版(CU毕业证)卡尔顿大学毕业证成绩单
一比一原版(CU毕业证)卡尔顿大学毕业证成绩单一比一原版(CU毕业证)卡尔顿大学毕业证成绩单
一比一原版(CU毕业证)卡尔顿大学毕业证成绩单yhkoc
 
Opendatabay - Open Data Marketplace.pptx
Opendatabay - Open Data Marketplace.pptxOpendatabay - Open Data Marketplace.pptx
Opendatabay - Open Data Marketplace.pptxOpendatabay
 
Empowering Data Analytics Ecosystem.pptx
Empowering Data Analytics Ecosystem.pptxEmpowering Data Analytics Ecosystem.pptx
Empowering Data Analytics Ecosystem.pptxbenishzehra469
 
Business update Q1 2024 Lar España Real Estate SOCIMI
Business update Q1 2024 Lar España Real Estate SOCIMIBusiness update Q1 2024 Lar España Real Estate SOCIMI
Business update Q1 2024 Lar España Real Estate SOCIMIAlejandraGmez176757
 

Recently uploaded (20)

Professional Data Engineer Certification Exam Guide  _  Learn  _  Google Clou...
Professional Data Engineer Certification Exam Guide  _  Learn  _  Google Clou...Professional Data Engineer Certification Exam Guide  _  Learn  _  Google Clou...
Professional Data Engineer Certification Exam Guide  _  Learn  _  Google Clou...
 
Jpolillo Amazon PPC - Bid Optimization Sample
Jpolillo Amazon PPC - Bid Optimization SampleJpolillo Amazon PPC - Bid Optimization Sample
Jpolillo Amazon PPC - Bid Optimization Sample
 
Computer Presentation.pptx ecommerce advantage s
Computer Presentation.pptx ecommerce advantage sComputer Presentation.pptx ecommerce advantage s
Computer Presentation.pptx ecommerce advantage s
 
一比一原版(CBU毕业证)不列颠海角大学毕业证成绩单
一比一原版(CBU毕业证)不列颠海角大学毕业证成绩单一比一原版(CBU毕业证)不列颠海角大学毕业证成绩单
一比一原版(CBU毕业证)不列颠海角大学毕业证成绩单
 
一比一原版(ArtEZ毕业证)ArtEZ艺术学院毕业证成绩单
一比一原版(ArtEZ毕业证)ArtEZ艺术学院毕业证成绩单一比一原版(ArtEZ毕业证)ArtEZ艺术学院毕业证成绩单
一比一原版(ArtEZ毕业证)ArtEZ艺术学院毕业证成绩单
 
一比一原版(RUG毕业证)格罗宁根大学毕业证成绩单
一比一原版(RUG毕业证)格罗宁根大学毕业证成绩单一比一原版(RUG毕业证)格罗宁根大学毕业证成绩单
一比一原版(RUG毕业证)格罗宁根大学毕业证成绩单
 
一比一原版(UPenn毕业证)宾夕法尼亚大学毕业证成绩单
一比一原版(UPenn毕业证)宾夕法尼亚大学毕业证成绩单一比一原版(UPenn毕业证)宾夕法尼亚大学毕业证成绩单
一比一原版(UPenn毕业证)宾夕法尼亚大学毕业证成绩单
 
tapal brand analysis PPT slide for comptetive data
tapal brand analysis PPT slide for comptetive datatapal brand analysis PPT slide for comptetive data
tapal brand analysis PPT slide for comptetive data
 
Algorithmic optimizations for Dynamic Levelwise PageRank (from STICD) : SHORT...
Algorithmic optimizations for Dynamic Levelwise PageRank (from STICD) : SHORT...Algorithmic optimizations for Dynamic Levelwise PageRank (from STICD) : SHORT...
Algorithmic optimizations for Dynamic Levelwise PageRank (from STICD) : SHORT...
 
Introduction-to-Cybersecurit57hhfcbbcxxx
Introduction-to-Cybersecurit57hhfcbbcxxxIntroduction-to-Cybersecurit57hhfcbbcxxx
Introduction-to-Cybersecurit57hhfcbbcxxx
 
Using PDB Relocation to Move a Single PDB to Another Existing CDB
Using PDB Relocation to Move a Single PDB to Another Existing CDBUsing PDB Relocation to Move a Single PDB to Another Existing CDB
Using PDB Relocation to Move a Single PDB to Another Existing CDB
 
Investigate & Recover / StarCompliance.io / Crypto_Crimes
Investigate & Recover / StarCompliance.io / Crypto_CrimesInvestigate & Recover / StarCompliance.io / Crypto_Crimes
Investigate & Recover / StarCompliance.io / Crypto_Crimes
 
哪里卖(usq毕业证书)南昆士兰大学毕业证研究生文凭证书托福证书原版一模一样
哪里卖(usq毕业证书)南昆士兰大学毕业证研究生文凭证书托福证书原版一模一样哪里卖(usq毕业证书)南昆士兰大学毕业证研究生文凭证书托福证书原版一模一样
哪里卖(usq毕业证书)南昆士兰大学毕业证研究生文凭证书托福证书原版一模一样
 
Tabula.io Cheatsheet: automate your data workflows
Tabula.io Cheatsheet: automate your data workflowsTabula.io Cheatsheet: automate your data workflows
Tabula.io Cheatsheet: automate your data workflows
 
一比一原版(CBU毕业证)卡普顿大学毕业证成绩单
一比一原版(CBU毕业证)卡普顿大学毕业证成绩单一比一原版(CBU毕业证)卡普顿大学毕业证成绩单
一比一原版(CBU毕业证)卡普顿大学毕业证成绩单
 
做(mqu毕业证书)麦考瑞大学毕业证硕士文凭证书学费发票原版一模一样
做(mqu毕业证书)麦考瑞大学毕业证硕士文凭证书学费发票原版一模一样做(mqu毕业证书)麦考瑞大学毕业证硕士文凭证书学费发票原版一模一样
做(mqu毕业证书)麦考瑞大学毕业证硕士文凭证书学费发票原版一模一样
 
一比一原版(CU毕业证)卡尔顿大学毕业证成绩单
一比一原版(CU毕业证)卡尔顿大学毕业证成绩单一比一原版(CU毕业证)卡尔顿大学毕业证成绩单
一比一原版(CU毕业证)卡尔顿大学毕业证成绩单
 
Opendatabay - Open Data Marketplace.pptx
Opendatabay - Open Data Marketplace.pptxOpendatabay - Open Data Marketplace.pptx
Opendatabay - Open Data Marketplace.pptx
 
Empowering Data Analytics Ecosystem.pptx
Empowering Data Analytics Ecosystem.pptxEmpowering Data Analytics Ecosystem.pptx
Empowering Data Analytics Ecosystem.pptx
 
Business update Q1 2024 Lar España Real Estate SOCIMI
Business update Q1 2024 Lar España Real Estate SOCIMIBusiness update Q1 2024 Lar España Real Estate SOCIMI
Business update Q1 2024 Lar España Real Estate SOCIMI
 

Featured

Everything You Need To Know About ChatGPT
Everything You Need To Know About ChatGPTEverything You Need To Know About ChatGPT
Everything You Need To Know About ChatGPTExpeed Software
 
Product Design Trends in 2024 | Teenage Engineerings
Product Design Trends in 2024 | Teenage EngineeringsProduct Design Trends in 2024 | Teenage Engineerings
Product Design Trends in 2024 | Teenage EngineeringsPixeldarts
 
How Race, Age and Gender Shape Attitudes Towards Mental Health
How Race, Age and Gender Shape Attitudes Towards Mental HealthHow Race, Age and Gender Shape Attitudes Towards Mental Health
How Race, Age and Gender Shape Attitudes Towards Mental HealthThinkNow
 
AI Trends in Creative Operations 2024 by Artwork Flow.pdf
AI Trends in Creative Operations 2024 by Artwork Flow.pdfAI Trends in Creative Operations 2024 by Artwork Flow.pdf
AI Trends in Creative Operations 2024 by Artwork Flow.pdfmarketingartwork
 
PEPSICO Presentation to CAGNY Conference Feb 2024
PEPSICO Presentation to CAGNY Conference Feb 2024PEPSICO Presentation to CAGNY Conference Feb 2024
PEPSICO Presentation to CAGNY Conference Feb 2024Neil Kimberley
 
Content Methodology: A Best Practices Report (Webinar)
Content Methodology: A Best Practices Report (Webinar)Content Methodology: A Best Practices Report (Webinar)
Content Methodology: A Best Practices Report (Webinar)contently
 
How to Prepare For a Successful Job Search for 2024
How to Prepare For a Successful Job Search for 2024How to Prepare For a Successful Job Search for 2024
How to Prepare For a Successful Job Search for 2024Albert Qian
 
Social Media Marketing Trends 2024 // The Global Indie Insights
Social Media Marketing Trends 2024 // The Global Indie InsightsSocial Media Marketing Trends 2024 // The Global Indie Insights
Social Media Marketing Trends 2024 // The Global Indie InsightsKurio // The Social Media Age(ncy)
 
Trends In Paid Search: Navigating The Digital Landscape In 2024
Trends In Paid Search: Navigating The Digital Landscape In 2024Trends In Paid Search: Navigating The Digital Landscape In 2024
Trends In Paid Search: Navigating The Digital Landscape In 2024Search Engine Journal
 
5 Public speaking tips from TED - Visualized summary
5 Public speaking tips from TED - Visualized summary5 Public speaking tips from TED - Visualized summary
5 Public speaking tips from TED - Visualized summarySpeakerHub
 
ChatGPT and the Future of Work - Clark Boyd
ChatGPT and the Future of Work - Clark Boyd ChatGPT and the Future of Work - Clark Boyd
ChatGPT and the Future of Work - Clark Boyd Clark Boyd
 
Getting into the tech field. what next
Getting into the tech field. what next Getting into the tech field. what next
Getting into the tech field. what next Tessa Mero
 
Google's Just Not That Into You: Understanding Core Updates & Search Intent
Google's Just Not That Into You: Understanding Core Updates & Search IntentGoogle's Just Not That Into You: Understanding Core Updates & Search Intent
Google's Just Not That Into You: Understanding Core Updates & Search IntentLily Ray
 
Time Management & Productivity - Best Practices
Time Management & Productivity -  Best PracticesTime Management & Productivity -  Best Practices
Time Management & Productivity - Best PracticesVit Horky
 
The six step guide to practical project management
The six step guide to practical project managementThe six step guide to practical project management
The six step guide to practical project managementMindGenius
 
Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...
Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...
Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...RachelPearson36
 
Unlocking the Power of ChatGPT and AI in Testing - A Real-World Look, present...
Unlocking the Power of ChatGPT and AI in Testing - A Real-World Look, present...Unlocking the Power of ChatGPT and AI in Testing - A Real-World Look, present...
Unlocking the Power of ChatGPT and AI in Testing - A Real-World Look, present...Applitools
 

Featured (20)

Everything You Need To Know About ChatGPT
Everything You Need To Know About ChatGPTEverything You Need To Know About ChatGPT
Everything You Need To Know About ChatGPT
 
Product Design Trends in 2024 | Teenage Engineerings
Product Design Trends in 2024 | Teenage EngineeringsProduct Design Trends in 2024 | Teenage Engineerings
Product Design Trends in 2024 | Teenage Engineerings
 
How Race, Age and Gender Shape Attitudes Towards Mental Health
How Race, Age and Gender Shape Attitudes Towards Mental HealthHow Race, Age and Gender Shape Attitudes Towards Mental Health
How Race, Age and Gender Shape Attitudes Towards Mental Health
 
AI Trends in Creative Operations 2024 by Artwork Flow.pdf
AI Trends in Creative Operations 2024 by Artwork Flow.pdfAI Trends in Creative Operations 2024 by Artwork Flow.pdf
AI Trends in Creative Operations 2024 by Artwork Flow.pdf
 
Skeleton Culture Code
Skeleton Culture CodeSkeleton Culture Code
Skeleton Culture Code
 
PEPSICO Presentation to CAGNY Conference Feb 2024
PEPSICO Presentation to CAGNY Conference Feb 2024PEPSICO Presentation to CAGNY Conference Feb 2024
PEPSICO Presentation to CAGNY Conference Feb 2024
 
Content Methodology: A Best Practices Report (Webinar)
Content Methodology: A Best Practices Report (Webinar)Content Methodology: A Best Practices Report (Webinar)
Content Methodology: A Best Practices Report (Webinar)
 
How to Prepare For a Successful Job Search for 2024
How to Prepare For a Successful Job Search for 2024How to Prepare For a Successful Job Search for 2024
How to Prepare For a Successful Job Search for 2024
 
Social Media Marketing Trends 2024 // The Global Indie Insights
Social Media Marketing Trends 2024 // The Global Indie InsightsSocial Media Marketing Trends 2024 // The Global Indie Insights
Social Media Marketing Trends 2024 // The Global Indie Insights
 
Trends In Paid Search: Navigating The Digital Landscape In 2024
Trends In Paid Search: Navigating The Digital Landscape In 2024Trends In Paid Search: Navigating The Digital Landscape In 2024
Trends In Paid Search: Navigating The Digital Landscape In 2024
 
5 Public speaking tips from TED - Visualized summary
5 Public speaking tips from TED - Visualized summary5 Public speaking tips from TED - Visualized summary
5 Public speaking tips from TED - Visualized summary
 
ChatGPT and the Future of Work - Clark Boyd
ChatGPT and the Future of Work - Clark Boyd ChatGPT and the Future of Work - Clark Boyd
ChatGPT and the Future of Work - Clark Boyd
 
Getting into the tech field. what next
Getting into the tech field. what next Getting into the tech field. what next
Getting into the tech field. what next
 
Google's Just Not That Into You: Understanding Core Updates & Search Intent
Google's Just Not That Into You: Understanding Core Updates & Search IntentGoogle's Just Not That Into You: Understanding Core Updates & Search Intent
Google's Just Not That Into You: Understanding Core Updates & Search Intent
 
How to have difficult conversations
How to have difficult conversations How to have difficult conversations
How to have difficult conversations
 
Introduction to Data Science
Introduction to Data ScienceIntroduction to Data Science
Introduction to Data Science
 
Time Management & Productivity - Best Practices
Time Management & Productivity -  Best PracticesTime Management & Productivity -  Best Practices
Time Management & Productivity - Best Practices
 
The six step guide to practical project management
The six step guide to practical project managementThe six step guide to practical project management
The six step guide to practical project management
 
Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...
Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...
Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...
 
Unlocking the Power of ChatGPT and AI in Testing - A Real-World Look, present...
Unlocking the Power of ChatGPT and AI in Testing - A Real-World Look, present...Unlocking the Power of ChatGPT and AI in Testing - A Real-World Look, present...
Unlocking the Power of ChatGPT and AI in Testing - A Real-World Look, present...
 

Image Generation with Tensorflow

  • 1. Image Generation with Tensorflow Cenk Bircano˘glu Boyner Group/Bah¸ce¸sehir Uni cenk.bircanoglu@gmail.com April 21, 2018 Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 1 / 44
  • 2. Overview 1 Convolutional Neural Networks 2 What are Generative Adverserial Networks? 3 Common GAN Architectures and Applications 4 Implementing GANs with Tensorflow 5 Question&Answer Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 2 / 44
  • 3. Convolutional Neural Networks Source: https://www.mathworks.com/discovery/convolutional-neural-network.html Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 3 / 44
  • 4. Deceiving the CNN Source: https://blog.openai.com/adversarial-example-research/ Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 4 / 44
  • 5. What are Generative Adverserial Networks? Source: https://deeplearning4j.org/generative-adversarial-network Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 5 / 44
  • 6. GAN Training Source: Goodfellow, 20017 Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 6 / 44
  • 7. Generator Source: http://arxiv.org/abs/1511.06434 Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 7 / 44
  • 8. Disriminator Source: http://arxiv.org/abs/1511.06434 Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 8 / 44
  • 9. Common GAN Applications Generating High-Quality Images Semi-supervised learning Ensembles Clustering Image blending Image Inpainting Joint Probability Super-Resolution De-occlusion Improving Classification & Recognition Semantic Segmentation Object Detection RNN-GANs Conditional Adversarial Nets Video Prediction & Generation Texture Synthesis & Style Transfer Image Translation GAN Theory 3-Dimensional GANs Music Face Generation & Editing Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 9 / 44
  • 10. GAN Results Source: https://kaonashi-tyc.github.io/2017/04/06/zi2zi.html, https://github.com/hanzhanggit/StackGAN, https://arxiv.org/pdf/1708.05509.pdf, https://arxiv.org/pdf/1705.09368.pdf, https://github.com/david-gpu/srez, http://research.nvidia.com/sites/default/files/pubs/2017-10 Progressive-Growing-of/karras2018iclr-paper.pdf, https://github.com/Guim3/IcGAN Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 10 / 44
  • 11. Generative Adversarial Networks - 2014 The first paper Two-player game that the generator is trained to generate images from inputed noises to fool the discriminator while the discriminator is trained to well discriminate real samples and fake samples E real(log(D)) + E fake(log(1-D)) https://papers.nips.cc/paper/5423-generative-adversarial-nets.pdf Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 11 / 44
  • 12. Conditional Generative Adversarial Nets - 2014 Source: http://guimperarnau.com/blog/2017/03/Fantastic-GANs-and-where-to-find-them https://arxiv.org/pdf/1511.06434.pdf Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 12 / 44
  • 13. cGAN Results Source: https://arxiv.org/pdf/1411.1784.pdf Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 13 / 44
  • 14. Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks - 2015 Source: https://arxiv.org/pdf/1511.06434.pdf https://arxiv.org/pdf/1511.06434.pdfCenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 14 / 44
  • 15. StackGAN: Text to Photo-realistic Image Synthesis with Stacked Generative Adversarial Networks - 2016 Source: https://arxiv.org/pdf/1612.03242.pdf https://arxiv.org/pdf/1612.03242.pdf Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 15 / 44
  • 16. StackGAN Results Source: https://arxiv.org/pdf/1612.03242.pdf Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 16 / 44
  • 17. InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets - 2016 Source: https://towardsdatascience.com/infogan-generative-adversarial-networks-part-iii-380c0c6712cd https://arxiv.org/pdf/1606.03657.pdf Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 17 / 44
  • 18. InfoGAN Results Source: https://arxiv.org/pdf/1606.03657.pdf Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 18 / 44
  • 19. Wasserstein GAN - 2017 Source: https://arxiv.org/pdf/1701.07875.pdf https://arxiv.org/pdf/1701.07875.pdf Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 19 / 44
  • 20. Improved Training of Wasserstein GANs - 2017 Source: https://arxiv.org/pdf/1704.00028.pdf https://arxiv.org/pdf/1704.00028.pdf Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 20 / 44
  • 21. WGAN-GP Results - 2017 Source: https://arxiv.org/pdf/1704.00028.pdf Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 21 / 44
  • 22. BEGAN: Boundary Equilibrium Generative Adversarial Networks - 2017 Source: https://arxiv.org/pdf/1703.10717.pdf https://arxiv.org/pdf/1703.10717.pdf Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 22 / 44
  • 23. BEGAN Results Source: https://arxiv.org/pdf/1703.10717.pdf Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 23 / 44
  • 24. Progressive Growing of GANs for Improved Quality, Stability, and Variation - 2017 Source: https://arxiv.org/pdf/1710.10196.pdf https://arxiv.org/pdf/1710.10196.pdf Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 24 / 44
  • 25. ProGAN Results Source: https://arxiv.org/pdf/1710.10196.pdf https://arxiv.org/pdf/1710.10196.pdf Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 25 / 44
  • 26. Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks - 2017 Source: https://arxiv.org/pdf/1703.10593.pdf https://arxiv.org/pdf/1703.10593.pdf Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 26 / 44
  • 27. CycleGAN Results Source: https://arxiv.org/pdf/1703.10593.pdf Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 27 / 44
  • 28. Implementing GANs with Tensorflow Generator Network Discriminator Network Initialize Graph Loss, Optimizer, Weights Training Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 28 / 44
  • 29. Discriminator in Tensorflow Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 29 / 44
  • 30. First Convolutional Layer in Discriminator Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 30 / 44
  • 31. Second Convolutional Layer in Discriminator Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 31 / 44
  • 32. First Fully Connected Layer in Discriminator Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 32 / 44
  • 33. Second Fully Connected Layer in Discriminator Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 33 / 44
  • 34. Generator in Tensorflow Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 34 / 44
  • 35. First Convolutional Layer in Generator Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 35 / 44
  • 36. Second Convolutional Layer in Generator Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 36 / 44
  • 37. Third Convolutional Layer in Generator Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 37 / 44
  • 38. Fourth Convolutional Layer in Generator Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 38 / 44
  • 39. Initialize Graph Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 39 / 44
  • 40. Loss, Optimizer, Weights in Tensorflow Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 40 / 44
  • 41. Training Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 41 / 44
  • 42. Results Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 42 / 44
  • 43. GAN Training Tips & Tricks Normalize the inputs A modified loss function Use a spherical Z BatchNorm Avoid Sparse Gradients: ReLU, MaxPool Use Soft and Noisy Labels DCGAN / Hybrid Models Use stability tricks from RL Use the ADAM Optimizer Track failures early Dont balance loss via statistics If you have labels, use them Add noise to inputs, decay over time Train discriminator more Batch Discrimination Discrete variables in Conditional GANs Use Dropouts in G in both train and test phase https://github.com/soumith/ganhacks Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 43 / 44
  • 44. Cenk Bircano˘glu (Boyner Group/Bah¸ce¸sehir Uni) GANs April 21, 2018 44 / 44