SlideShare a Scribd company logo
1 of 23
InfoGAN
Interpretable Representation Learning by Information
Maximizing Generative Adversarial Nets
Introduction
 In ordinary GANs, the latent vector z has an arbitrary
representation. The vector has no intrinsic meaning.
 It would be desirable to make a more meaningful
representation of the outputs.
Disentangled Representation
 We wish to disentangle the representation of the output
images in the input latent vectors.
 That is, we wish to make the values of the latent vector c
correspond to features in the generated images.
Information Theory
A brief introduction to relevant concepts in information theory.
Entropy
Entropy can be intuitively
understood the amount of
information that a variable contains.
It is borrowed from statistical
thermodynamics and is directly
analogous to the physical definition
of randomness of particle states.
𝐻 𝑋 = −
𝑖=1
𝑛
𝑃 𝑥𝑖 log 𝑏 𝑃 𝑥𝑖
𝐻 𝑋 𝑌 = −
𝑖,𝑗
𝑝 𝑥𝑖, 𝑦𝑗 𝑙𝑜𝑔
𝑝(𝑥𝑖, 𝑦𝑖)
𝑝 𝑦𝑖
Mutual Information: Definition
 In information theory, I(X;Y),
the mutual information
between X and Y, measures
the “amount of information”
learned from knowledge of
random variable Y about the
random variable X.
 The mutual information can be
expressed as the difference of
two entropy terms:
I(X;Y) = H(X)-H(X|Y) = H(Y)-H(Y|X)
Mutual Information: Interpretation
 If X and Y are independent,
then I(X;Y) = 0, because
knowing one variable reveals
nothing about the other.
 If X and Y are related by a
deterministic, invertible
function, I(X;Y) is at its
maximum.
 I(X;Y) is the reduction of
uncertainty in X when Y is
observed and vice versa.
Mutual Information: Implications
 Formulation of a cost function using mutual information.
 Information regularized mini-max game.
min
𝐺
max
𝐷
𝑉𝐼 𝐷, 𝐺 = 𝑉 𝐷, 𝐺 − 𝜆𝐼(𝑐; 𝐺 𝑧, 𝑐 )
z: vector for representing intrinsic noise
c: Latent code representing meaningful information.
Implementation
Variational Mutual Information Maximization and Approximations.
Variational Mutual Information Maximization
 In practice, 𝐼(𝑐; 𝐺 𝑧, 𝑐 ) cannot be maximized directly as this
requires 𝑃(𝑐|𝑥), the posterior distribution.
 We can obtain a lower bound for 𝐼(𝑐; 𝐺 𝑧, 𝑐 ) by using an
auxiliary distribution, 𝑄 𝑐 𝑥 , to approximate 𝑃(𝑐|𝑥).
 This technique is known as Variational Mutual Information
Maximization. The equations are in the next slide.
Variational Mutual Information Maximization
 The Mutual Information is decomposed into its components.
 Q is introduced to use the definition of KL divergence.
 KL Divergence is always greater than 0 by definition.
Variational Mutual Information Maximization
 Although 𝐻(𝑐) can also be optimized, it is set as a constant for
simplicity. This is done by drawing c from a fixed distribution.
 The equality in this part is proven in the appendix of the paper. It
holds under most conditions.
Variational Mutual Information Maximization
 By the proofs discussed previously, the actual information
regularization is as follows.
min
𝐺,𝑄
max
𝐷
𝑉𝐼𝑛𝑓𝑜𝐺𝐴𝑁 𝐷, 𝐺, 𝑄 = 𝑉 𝐷, 𝐺 − 𝜆𝐿𝐼(𝐺, 𝑄)
𝐿𝐼: Information Lower Bound
𝜆: Weighting Hyperparameter
Practical Implementation
 In practice, we use a neural
network to represent Q.
 KL Divergence is minimized
when Q converges to P.
 Q is just D with an extra FC
layer. It outputs an estimate
of the c latent vector(s).
 𝐿𝐼(𝐺, 𝑄) has been observed
to converge faster than GAN
objectives.
 InfoGAN thus comes
essentially for free with GAN.
Overview of the Architecture
Additional explanation
 Cross Entropy Loss is used on the estimate of c and the actual c
(Some implementations use MSE).
 For discrete cases, the outputs are softmax activations. For
continuous cases, they may be sigmoid, tanh, or linear.
 The original is much more complicated and does not directly
output the estimate of the latent vector but estimates of its
distribution (mean and standard deviation).
Analysis
 The information lower bound
𝐿𝐼 does not increase without
information regularization.
 The maximum appears to be
2.3 for this case.
Results: MNIST
The discrete latent vector input for digits is
highly correlated with the output number.
It can be used to classify MNIST with a 5%
error rate, even when InfoGAN is trained
without any labels.
The continuous latent vector has found
the angle of the characters.
We confirm that the latent code validity by
extending the range beyond the original.
The ordinary GAN has learned nothing.
There is no control over which vector
learns what.
Results: Faces
Multiple vectors with a range between -1
and 1 are used.
The 4 latent vectors appear to have
captured azimuth, elevation, lighting, and
width.
There is smooth interpolation within and
even beyond the range.
Moreover, the other details change to
make a much more natural image.
It is not a simple case where only the
target feature changes while the other
factors remain unnaturally constant.
Results on chairs
Results on SVHN
Results on CelebA
Conclusion
 InfoGAN can learn and interpret salient features of data
without any labels or supervision.
 InfoGAN can discover salient latent factors of variation
automatically and produce a latent vector containing that
information.

More Related Content

What's hot

[DL輪読会]Graph Convolutional Policy Network for Goal-Directed Molecular Graph G...
[DL輪読会]Graph Convolutional Policy Network for Goal-Directed Molecular Graph G...[DL輪読会]Graph Convolutional Policy Network for Goal-Directed Molecular Graph G...
[DL輪読会]Graph Convolutional Policy Network for Goal-Directed Molecular Graph G...Deep Learning JP
 
PR-409: Denoising Diffusion Probabilistic Models
PR-409: Denoising Diffusion Probabilistic ModelsPR-409: Denoising Diffusion Probabilistic Models
PR-409: Denoising Diffusion Probabilistic ModelsHyeongmin Lee
 
[DL輪読会]Scalable Training of Inference Networks for Gaussian-Process Models
[DL輪読会]Scalable Training of Inference Networks for Gaussian-Process Models[DL輪読会]Scalable Training of Inference Networks for Gaussian-Process Models
[DL輪読会]Scalable Training of Inference Networks for Gaussian-Process ModelsDeep Learning JP
 
[DL輪読会]SOM-VAE: Interpretable Discrete Representation Learning on Time Series
[DL輪読会]SOM-VAE: Interpretable Discrete Representation Learning on Time Series[DL輪読会]SOM-VAE: Interpretable Discrete Representation Learning on Time Series
[DL輪読会]SOM-VAE: Interpretable Discrete Representation Learning on Time SeriesDeep Learning JP
 
Diffusion models beat gans on image synthesis
Diffusion models beat gans on image synthesisDiffusion models beat gans on image synthesis
Diffusion models beat gans on image synthesisBeerenSahu
 
InfoGAN: Interpretable Representation Learning by Information Maximizing Gene...
InfoGAN: Interpretable Representation Learning by Information Maximizing Gene...InfoGAN: Interpretable Representation Learning by Information Maximizing Gene...
InfoGAN: Interpretable Representation Learning by Information Maximizing Gene...홍배 김
 
【論文読み会】Self-Attention Generative Adversarial Networks
【論文読み会】Self-Attention Generative  Adversarial Networks【論文読み会】Self-Attention Generative  Adversarial Networks
【論文読み会】Self-Attention Generative Adversarial NetworksARISE analytics
 
Reducing the Dimensionality of Data with Neural Networks
Reducing the Dimensionality of Data with Neural NetworksReducing the Dimensionality of Data with Neural Networks
Reducing the Dimensionality of Data with Neural NetworksNagayoshi Yamashita
 
Dimensionality reduction with UMAP
Dimensionality reduction with UMAPDimensionality reduction with UMAP
Dimensionality reduction with UMAPJakub Bartczuk
 
Image-to-Image Translation
Image-to-Image TranslationImage-to-Image Translation
Image-to-Image TranslationJunho Kim
 
is anyone_interest_in_auto-encoding_variational-bayes
is anyone_interest_in_auto-encoding_variational-bayesis anyone_interest_in_auto-encoding_variational-bayes
is anyone_interest_in_auto-encoding_variational-bayesNAVER Engineering
 
ML - Multiple Linear Regression
ML - Multiple Linear RegressionML - Multiple Linear Regression
ML - Multiple Linear RegressionAndrew Ferlitsch
 
Variational Autoencoder Tutorial
Variational Autoencoder Tutorial Variational Autoencoder Tutorial
Variational Autoencoder Tutorial Hojin Yang
 
Deep generative model.pdf
Deep generative model.pdfDeep generative model.pdf
Deep generative model.pdfHyungjoo Cho
 
Unsupervised learning represenation with DCGAN
Unsupervised learning represenation with DCGANUnsupervised learning represenation with DCGAN
Unsupervised learning represenation with DCGANShyam Krishna Khadka
 
[DL輪読会]"CyCADA: Cycle-Consistent Adversarial Domain Adaptation"&"Learning Se...
 [DL輪読会]"CyCADA: Cycle-Consistent Adversarial Domain Adaptation"&"Learning Se... [DL輪読会]"CyCADA: Cycle-Consistent Adversarial Domain Adaptation"&"Learning Se...
[DL輪読会]"CyCADA: Cycle-Consistent Adversarial Domain Adaptation"&"Learning Se...Deep Learning JP
 
深層生成モデルを用いたマルチモーダル学習
深層生成モデルを用いたマルチモーダル学習深層生成モデルを用いたマルチモーダル学習
深層生成モデルを用いたマルチモーダル学習Masahiro Suzuki
 
[DL輪読会]The Neural Process Family−Neural Processes関連の実装を読んで動かしてみる−
[DL輪読会]The Neural Process Family−Neural Processes関連の実装を読んで動かしてみる−[DL輪読会]The Neural Process Family−Neural Processes関連の実装を読んで動かしてみる−
[DL輪読会]The Neural Process Family−Neural Processes関連の実装を読んで動かしてみる−Deep Learning JP
 

What's hot (20)

[DL輪読会]Graph Convolutional Policy Network for Goal-Directed Molecular Graph G...
[DL輪読会]Graph Convolutional Policy Network for Goal-Directed Molecular Graph G...[DL輪読会]Graph Convolutional Policy Network for Goal-Directed Molecular Graph G...
[DL輪読会]Graph Convolutional Policy Network for Goal-Directed Molecular Graph G...
 
PR-409: Denoising Diffusion Probabilistic Models
PR-409: Denoising Diffusion Probabilistic ModelsPR-409: Denoising Diffusion Probabilistic Models
PR-409: Denoising Diffusion Probabilistic Models
 
[DL輪読会]Scalable Training of Inference Networks for Gaussian-Process Models
[DL輪読会]Scalable Training of Inference Networks for Gaussian-Process Models[DL輪読会]Scalable Training of Inference Networks for Gaussian-Process Models
[DL輪読会]Scalable Training of Inference Networks for Gaussian-Process Models
 
Activation function
Activation functionActivation function
Activation function
 
[DL輪読会]SOM-VAE: Interpretable Discrete Representation Learning on Time Series
[DL輪読会]SOM-VAE: Interpretable Discrete Representation Learning on Time Series[DL輪読会]SOM-VAE: Interpretable Discrete Representation Learning on Time Series
[DL輪読会]SOM-VAE: Interpretable Discrete Representation Learning on Time Series
 
Diffusion models beat gans on image synthesis
Diffusion models beat gans on image synthesisDiffusion models beat gans on image synthesis
Diffusion models beat gans on image synthesis
 
InfoGAN: Interpretable Representation Learning by Information Maximizing Gene...
InfoGAN: Interpretable Representation Learning by Information Maximizing Gene...InfoGAN: Interpretable Representation Learning by Information Maximizing Gene...
InfoGAN: Interpretable Representation Learning by Information Maximizing Gene...
 
【論文読み会】Self-Attention Generative Adversarial Networks
【論文読み会】Self-Attention Generative  Adversarial Networks【論文読み会】Self-Attention Generative  Adversarial Networks
【論文読み会】Self-Attention Generative Adversarial Networks
 
Reducing the Dimensionality of Data with Neural Networks
Reducing the Dimensionality of Data with Neural NetworksReducing the Dimensionality of Data with Neural Networks
Reducing the Dimensionality of Data with Neural Networks
 
Linear algebra
Linear algebraLinear algebra
Linear algebra
 
Dimensionality reduction with UMAP
Dimensionality reduction with UMAPDimensionality reduction with UMAP
Dimensionality reduction with UMAP
 
Image-to-Image Translation
Image-to-Image TranslationImage-to-Image Translation
Image-to-Image Translation
 
is anyone_interest_in_auto-encoding_variational-bayes
is anyone_interest_in_auto-encoding_variational-bayesis anyone_interest_in_auto-encoding_variational-bayes
is anyone_interest_in_auto-encoding_variational-bayes
 
ML - Multiple Linear Regression
ML - Multiple Linear RegressionML - Multiple Linear Regression
ML - Multiple Linear Regression
 
Variational Autoencoder Tutorial
Variational Autoencoder Tutorial Variational Autoencoder Tutorial
Variational Autoencoder Tutorial
 
Deep generative model.pdf
Deep generative model.pdfDeep generative model.pdf
Deep generative model.pdf
 
Unsupervised learning represenation with DCGAN
Unsupervised learning represenation with DCGANUnsupervised learning represenation with DCGAN
Unsupervised learning represenation with DCGAN
 
[DL輪読会]"CyCADA: Cycle-Consistent Adversarial Domain Adaptation"&"Learning Se...
 [DL輪読会]"CyCADA: Cycle-Consistent Adversarial Domain Adaptation"&"Learning Se... [DL輪読会]"CyCADA: Cycle-Consistent Adversarial Domain Adaptation"&"Learning Se...
[DL輪読会]"CyCADA: Cycle-Consistent Adversarial Domain Adaptation"&"Learning Se...
 
深層生成モデルを用いたマルチモーダル学習
深層生成モデルを用いたマルチモーダル学習深層生成モデルを用いたマルチモーダル学習
深層生成モデルを用いたマルチモーダル学習
 
[DL輪読会]The Neural Process Family−Neural Processes関連の実装を読んで動かしてみる−
[DL輪読会]The Neural Process Family−Neural Processes関連の実装を読んで動かしてみる−[DL輪読会]The Neural Process Family−Neural Processes関連の実装を読んで動かしてみる−
[DL輪読会]The Neural Process Family−Neural Processes関連の実装を読んで動かしてみる−
 

Similar to Interpretable Representation Learning with InfoGAN

An Image representation using Compressive Sensing and Arithmetic Coding
An Image representation using Compressive Sensing and Arithmetic Coding   An Image representation using Compressive Sensing and Arithmetic Coding
An Image representation using Compressive Sensing and Arithmetic Coding IJCERT
 
Deep learning MindMap
Deep learning MindMapDeep learning MindMap
Deep learning MindMapAshish Patel
 
V. pacáková, d. brebera
V. pacáková, d. breberaV. pacáková, d. brebera
V. pacáková, d. breberalogyalaa
 
Artificial intelligence
Artificial intelligenceArtificial intelligence
Artificial intelligencekeerthikaA8
 
Artificial intelligence.pptx
Artificial intelligence.pptxArtificial intelligence.pptx
Artificial intelligence.pptxkeerthikaA8
 
Artificial intelligence
Artificial intelligenceArtificial intelligence
Artificial intelligencekeerthikaA8
 
Quantitative Propagation of Chaos for SGD in Wide Neural Networks
Quantitative Propagation of Chaos for SGD in Wide Neural NetworksQuantitative Propagation of Chaos for SGD in Wide Neural Networks
Quantitative Propagation of Chaos for SGD in Wide Neural NetworksValentin De Bortoli
 
Classifiers
ClassifiersClassifiers
ClassifiersAyurdata
 
Descriptive_Statistics : Introduction to Descriptive_Statistics,Central tende...
Descriptive_Statistics : Introduction to Descriptive_Statistics,Central tende...Descriptive_Statistics : Introduction to Descriptive_Statistics,Central tende...
Descriptive_Statistics : Introduction to Descriptive_Statistics,Central tende...SanjanaSaxena17
 
Machine learning session4(linear regression)
Machine learning   session4(linear regression)Machine learning   session4(linear regression)
Machine learning session4(linear regression)Abhimanyu Dwivedi
 
Data Mining for Integration and Verification of Socio-Geographical Trend Stat...
Data Mining for Integration and Verification of Socio-Geographical Trend Stat...Data Mining for Integration and Verification of Socio-Geographical Trend Stat...
Data Mining for Integration and Verification of Socio-Geographical Trend Stat...ijaia
 
Data Mining for Integration and Verification of Socio-Geographical Trend Stat...
Data Mining for Integration and Verification of Socio-Geographical Trend Stat...Data Mining for Integration and Verification of Socio-Geographical Trend Stat...
Data Mining for Integration and Verification of Socio-Geographical Trend Stat...gerogepatton
 
Illustrative Introductory Neural Networks
Illustrative Introductory Neural NetworksIllustrative Introductory Neural Networks
Illustrative Introductory Neural NetworksYasutoTamura1
 
International Journal of Engineering Research and Development (IJERD)
 International Journal of Engineering Research and Development (IJERD) International Journal of Engineering Research and Development (IJERD)
International Journal of Engineering Research and Development (IJERD)IJERD Editor
 
Non-Normally Distributed Errors In Regression Diagnostics.docx
Non-Normally Distributed Errors In Regression Diagnostics.docxNon-Normally Distributed Errors In Regression Diagnostics.docx
Non-Normally Distributed Errors In Regression Diagnostics.docxvannagoforth
 
PRML Chapter 5
PRML Chapter 5PRML Chapter 5
PRML Chapter 5Sunwoo Kim
 
nonlinear_rmt.pdf
nonlinear_rmt.pdfnonlinear_rmt.pdf
nonlinear_rmt.pdfGieTe
 
(研究会輪読) Weight Uncertainty in Neural Networks
(研究会輪読) Weight Uncertainty in Neural Networks(研究会輪読) Weight Uncertainty in Neural Networks
(研究会輪読) Weight Uncertainty in Neural NetworksMasahiro Suzuki
 
Behavior study of entropy in a digital image through an iterative algorithm
Behavior study of entropy in a digital image through an iterative algorithmBehavior study of entropy in a digital image through an iterative algorithm
Behavior study of entropy in a digital image through an iterative algorithmijscmcj
 
A simple framework for contrastive learning of visual representations
A simple framework for contrastive learning of visual representationsA simple framework for contrastive learning of visual representations
A simple framework for contrastive learning of visual representationsDevansh16
 

Similar to Interpretable Representation Learning with InfoGAN (20)

An Image representation using Compressive Sensing and Arithmetic Coding
An Image representation using Compressive Sensing and Arithmetic Coding   An Image representation using Compressive Sensing and Arithmetic Coding
An Image representation using Compressive Sensing and Arithmetic Coding
 
Deep learning MindMap
Deep learning MindMapDeep learning MindMap
Deep learning MindMap
 
V. pacáková, d. brebera
V. pacáková, d. breberaV. pacáková, d. brebera
V. pacáková, d. brebera
 
Artificial intelligence
Artificial intelligenceArtificial intelligence
Artificial intelligence
 
Artificial intelligence.pptx
Artificial intelligence.pptxArtificial intelligence.pptx
Artificial intelligence.pptx
 
Artificial intelligence
Artificial intelligenceArtificial intelligence
Artificial intelligence
 
Quantitative Propagation of Chaos for SGD in Wide Neural Networks
Quantitative Propagation of Chaos for SGD in Wide Neural NetworksQuantitative Propagation of Chaos for SGD in Wide Neural Networks
Quantitative Propagation of Chaos for SGD in Wide Neural Networks
 
Classifiers
ClassifiersClassifiers
Classifiers
 
Descriptive_Statistics : Introduction to Descriptive_Statistics,Central tende...
Descriptive_Statistics : Introduction to Descriptive_Statistics,Central tende...Descriptive_Statistics : Introduction to Descriptive_Statistics,Central tende...
Descriptive_Statistics : Introduction to Descriptive_Statistics,Central tende...
 
Machine learning session4(linear regression)
Machine learning   session4(linear regression)Machine learning   session4(linear regression)
Machine learning session4(linear regression)
 
Data Mining for Integration and Verification of Socio-Geographical Trend Stat...
Data Mining for Integration and Verification of Socio-Geographical Trend Stat...Data Mining for Integration and Verification of Socio-Geographical Trend Stat...
Data Mining for Integration and Verification of Socio-Geographical Trend Stat...
 
Data Mining for Integration and Verification of Socio-Geographical Trend Stat...
Data Mining for Integration and Verification of Socio-Geographical Trend Stat...Data Mining for Integration and Verification of Socio-Geographical Trend Stat...
Data Mining for Integration and Verification of Socio-Geographical Trend Stat...
 
Illustrative Introductory Neural Networks
Illustrative Introductory Neural NetworksIllustrative Introductory Neural Networks
Illustrative Introductory Neural Networks
 
International Journal of Engineering Research and Development (IJERD)
 International Journal of Engineering Research and Development (IJERD) International Journal of Engineering Research and Development (IJERD)
International Journal of Engineering Research and Development (IJERD)
 
Non-Normally Distributed Errors In Regression Diagnostics.docx
Non-Normally Distributed Errors In Regression Diagnostics.docxNon-Normally Distributed Errors In Regression Diagnostics.docx
Non-Normally Distributed Errors In Regression Diagnostics.docx
 
PRML Chapter 5
PRML Chapter 5PRML Chapter 5
PRML Chapter 5
 
nonlinear_rmt.pdf
nonlinear_rmt.pdfnonlinear_rmt.pdf
nonlinear_rmt.pdf
 
(研究会輪読) Weight Uncertainty in Neural Networks
(研究会輪読) Weight Uncertainty in Neural Networks(研究会輪読) Weight Uncertainty in Neural Networks
(研究会輪読) Weight Uncertainty in Neural Networks
 
Behavior study of entropy in a digital image through an iterative algorithm
Behavior study of entropy in a digital image through an iterative algorithmBehavior study of entropy in a digital image through an iterative algorithm
Behavior study of entropy in a digital image through an iterative algorithm
 
A simple framework for contrastive learning of visual representations
A simple framework for contrastive learning of visual representationsA simple framework for contrastive learning of visual representations
A simple framework for contrastive learning of visual representations
 

More from Joonhyung Lee

Rethinking Attention with Performers
Rethinking Attention with PerformersRethinking Attention with Performers
Rethinking Attention with PerformersJoonhyung Lee
 
Denoising Unpaired Low Dose CT Images with Self-Ensembled CycleGAN
Denoising Unpaired Low Dose CT Images with Self-Ensembled CycleGANDenoising Unpaired Low Dose CT Images with Self-Ensembled CycleGAN
Denoising Unpaired Low Dose CT Images with Self-Ensembled CycleGANJoonhyung Lee
 
Deep Learning Fast MRI Using Channel Attention in Magnitude Domain
Deep Learning Fast MRI Using Channel Attention in Magnitude DomainDeep Learning Fast MRI Using Channel Attention in Magnitude Domain
Deep Learning Fast MRI Using Channel Attention in Magnitude DomainJoonhyung Lee
 
CutMix: Regularization Strategy to Train Strong Classifiers with Localizable ...
CutMix: Regularization Strategy to Train Strong Classifiers with Localizable ...CutMix: Regularization Strategy to Train Strong Classifiers with Localizable ...
CutMix: Regularization Strategy to Train Strong Classifiers with Localizable ...Joonhyung Lee
 
AlphaZero: A General Reinforcement Learning Algorithm that Masters Chess, Sho...
AlphaZero: A General Reinforcement Learning Algorithm that Masters Chess, Sho...AlphaZero: A General Reinforcement Learning Algorithm that Masters Chess, Sho...
AlphaZero: A General Reinforcement Learning Algorithm that Masters Chess, Sho...Joonhyung Lee
 
Squeeze Excitation Networks, The simple idea that won the final ImageNet Chal...
Squeeze Excitation Networks, The simple idea that won the final ImageNet Chal...Squeeze Excitation Networks, The simple idea that won the final ImageNet Chal...
Squeeze Excitation Networks, The simple idea that won the final ImageNet Chal...Joonhyung Lee
 
DeepLab V3+: Encoder-Decoder with Atrous Separable Convolution for Semantic I...
DeepLab V3+: Encoder-Decoder with Atrous Separable Convolution for Semantic I...DeepLab V3+: Encoder-Decoder with Atrous Separable Convolution for Semantic I...
DeepLab V3+: Encoder-Decoder with Atrous Separable Convolution for Semantic I...Joonhyung Lee
 
AlphaGo Zero: Mastering the Game of Go Without Human Knowledge
AlphaGo Zero: Mastering the Game of Go Without Human KnowledgeAlphaGo Zero: Mastering the Game of Go Without Human Knowledge
AlphaGo Zero: Mastering the Game of Go Without Human KnowledgeJoonhyung Lee
 
Deep Learning in Bio-Medical Imaging
Deep Learning in Bio-Medical ImagingDeep Learning in Bio-Medical Imaging
Deep Learning in Bio-Medical ImagingJoonhyung Lee
 

More from Joonhyung Lee (11)

nnUNet
nnUNetnnUNet
nnUNet
 
Rethinking Attention with Performers
Rethinking Attention with PerformersRethinking Attention with Performers
Rethinking Attention with Performers
 
Denoising Unpaired Low Dose CT Images with Self-Ensembled CycleGAN
Denoising Unpaired Low Dose CT Images with Self-Ensembled CycleGANDenoising Unpaired Low Dose CT Images with Self-Ensembled CycleGAN
Denoising Unpaired Low Dose CT Images with Self-Ensembled CycleGAN
 
Deep Learning Fast MRI Using Channel Attention in Magnitude Domain
Deep Learning Fast MRI Using Channel Attention in Magnitude DomainDeep Learning Fast MRI Using Channel Attention in Magnitude Domain
Deep Learning Fast MRI Using Channel Attention in Magnitude Domain
 
CutMix: Regularization Strategy to Train Strong Classifiers with Localizable ...
CutMix: Regularization Strategy to Train Strong Classifiers with Localizable ...CutMix: Regularization Strategy to Train Strong Classifiers with Localizable ...
CutMix: Regularization Strategy to Train Strong Classifiers with Localizable ...
 
AlphaZero: A General Reinforcement Learning Algorithm that Masters Chess, Sho...
AlphaZero: A General Reinforcement Learning Algorithm that Masters Chess, Sho...AlphaZero: A General Reinforcement Learning Algorithm that Masters Chess, Sho...
AlphaZero: A General Reinforcement Learning Algorithm that Masters Chess, Sho...
 
Squeeze Excitation Networks, The simple idea that won the final ImageNet Chal...
Squeeze Excitation Networks, The simple idea that won the final ImageNet Chal...Squeeze Excitation Networks, The simple idea that won the final ImageNet Chal...
Squeeze Excitation Networks, The simple idea that won the final ImageNet Chal...
 
DeepLab V3+: Encoder-Decoder with Atrous Separable Convolution for Semantic I...
DeepLab V3+: Encoder-Decoder with Atrous Separable Convolution for Semantic I...DeepLab V3+: Encoder-Decoder with Atrous Separable Convolution for Semantic I...
DeepLab V3+: Encoder-Decoder with Atrous Separable Convolution for Semantic I...
 
AlphaGo Zero: Mastering the Game of Go Without Human Knowledge
AlphaGo Zero: Mastering the Game of Go Without Human KnowledgeAlphaGo Zero: Mastering the Game of Go Without Human Knowledge
AlphaGo Zero: Mastering the Game of Go Without Human Knowledge
 
StarGAN
StarGANStarGAN
StarGAN
 
Deep Learning in Bio-Medical Imaging
Deep Learning in Bio-Medical ImagingDeep Learning in Bio-Medical Imaging
Deep Learning in Bio-Medical Imaging
 

Recently uploaded

The Black hole shadow in Modified Gravity
The Black hole shadow in Modified GravityThe Black hole shadow in Modified Gravity
The Black hole shadow in Modified GravitySubhadipsau21168
 
Recombinant DNA technology( Transgenic plant and animal)
Recombinant DNA technology( Transgenic plant and animal)Recombinant DNA technology( Transgenic plant and animal)
Recombinant DNA technology( Transgenic plant and animal)DHURKADEVIBASKAR
 
Analytical Profile of Coleus Forskohlii | Forskolin .pdf
Analytical Profile of Coleus Forskohlii | Forskolin .pdfAnalytical Profile of Coleus Forskohlii | Forskolin .pdf
Analytical Profile of Coleus Forskohlii | Forskolin .pdfSwapnil Therkar
 
PossibleEoarcheanRecordsoftheGeomagneticFieldPreservedintheIsuaSupracrustalBe...
PossibleEoarcheanRecordsoftheGeomagneticFieldPreservedintheIsuaSupracrustalBe...PossibleEoarcheanRecordsoftheGeomagneticFieldPreservedintheIsuaSupracrustalBe...
PossibleEoarcheanRecordsoftheGeomagneticFieldPreservedintheIsuaSupracrustalBe...Sérgio Sacani
 
A relative description on Sonoporation.pdf
A relative description on Sonoporation.pdfA relative description on Sonoporation.pdf
A relative description on Sonoporation.pdfnehabiju2046
 
Hubble Asteroid Hunter III. Physical properties of newly found asteroids
Hubble Asteroid Hunter III. Physical properties of newly found asteroidsHubble Asteroid Hunter III. Physical properties of newly found asteroids
Hubble Asteroid Hunter III. Physical properties of newly found asteroidsSérgio Sacani
 
Artificial Intelligence In Microbiology by Dr. Prince C P
Artificial Intelligence In Microbiology by Dr. Prince C PArtificial Intelligence In Microbiology by Dr. Prince C P
Artificial Intelligence In Microbiology by Dr. Prince C PPRINCE C P
 
Disentangling the origin of chemical differences using GHOST
Disentangling the origin of chemical differences using GHOSTDisentangling the origin of chemical differences using GHOST
Disentangling the origin of chemical differences using GHOSTSérgio Sacani
 
Work, Energy and Power for class 10 ICSE Physics
Work, Energy and Power for class 10 ICSE PhysicsWork, Energy and Power for class 10 ICSE Physics
Work, Energy and Power for class 10 ICSE Physicsvishikhakeshava1
 
Luciferase in rDNA technology (biotechnology).pptx
Luciferase in rDNA technology (biotechnology).pptxLuciferase in rDNA technology (biotechnology).pptx
Luciferase in rDNA technology (biotechnology).pptxAleenaTreesaSaji
 
Is RISC-V ready for HPC workload? Maybe?
Is RISC-V ready for HPC workload? Maybe?Is RISC-V ready for HPC workload? Maybe?
Is RISC-V ready for HPC workload? Maybe?Patrick Diehl
 
Behavioral Disorder: Schizophrenia & it's Case Study.pdf
Behavioral Disorder: Schizophrenia & it's Case Study.pdfBehavioral Disorder: Schizophrenia & it's Case Study.pdf
Behavioral Disorder: Schizophrenia & it's Case Study.pdfSELF-EXPLANATORY
 
Biopesticide (2).pptx .This slides helps to know the different types of biop...
Biopesticide (2).pptx  .This slides helps to know the different types of biop...Biopesticide (2).pptx  .This slides helps to know the different types of biop...
Biopesticide (2).pptx .This slides helps to know the different types of biop...RohitNehra6
 
Ahmedabad Call Girls Service 9537192988 can satisfy every one of your dreams
Ahmedabad Call Girls Service 9537192988 can satisfy every one of your dreamsAhmedabad Call Girls Service 9537192988 can satisfy every one of your dreams
Ahmedabad Call Girls Service 9537192988 can satisfy every one of your dreamsoolala9823
 
Boyles law module in the grade 10 science
Boyles law module in the grade 10 scienceBoyles law module in the grade 10 science
Boyles law module in the grade 10 sciencefloriejanemacaya1
 
Natural Polymer Based Nanomaterials
Natural Polymer Based NanomaterialsNatural Polymer Based Nanomaterials
Natural Polymer Based NanomaterialsAArockiyaNisha
 
Animal Communication- Auditory and Visual.pptx
Animal Communication- Auditory and Visual.pptxAnimal Communication- Auditory and Visual.pptx
Animal Communication- Auditory and Visual.pptxUmerFayaz5
 
Dashanga agada a formulation of Agada tantra dealt in 3 Rd year bams agada tanta
Dashanga agada a formulation of Agada tantra dealt in 3 Rd year bams agada tantaDashanga agada a formulation of Agada tantra dealt in 3 Rd year bams agada tanta
Dashanga agada a formulation of Agada tantra dealt in 3 Rd year bams agada tantaPraksha3
 
SOLUBLE PATTERN RECOGNITION RECEPTORS.pptx
SOLUBLE PATTERN RECOGNITION RECEPTORS.pptxSOLUBLE PATTERN RECOGNITION RECEPTORS.pptx
SOLUBLE PATTERN RECOGNITION RECEPTORS.pptxkessiyaTpeter
 

Recently uploaded (20)

The Black hole shadow in Modified Gravity
The Black hole shadow in Modified GravityThe Black hole shadow in Modified Gravity
The Black hole shadow in Modified Gravity
 
Recombinant DNA technology( Transgenic plant and animal)
Recombinant DNA technology( Transgenic plant and animal)Recombinant DNA technology( Transgenic plant and animal)
Recombinant DNA technology( Transgenic plant and animal)
 
Analytical Profile of Coleus Forskohlii | Forskolin .pdf
Analytical Profile of Coleus Forskohlii | Forskolin .pdfAnalytical Profile of Coleus Forskohlii | Forskolin .pdf
Analytical Profile of Coleus Forskohlii | Forskolin .pdf
 
9953056974 Young Call Girls In Mahavir enclave Indian Quality Escort service
9953056974 Young Call Girls In Mahavir enclave Indian Quality Escort service9953056974 Young Call Girls In Mahavir enclave Indian Quality Escort service
9953056974 Young Call Girls In Mahavir enclave Indian Quality Escort service
 
PossibleEoarcheanRecordsoftheGeomagneticFieldPreservedintheIsuaSupracrustalBe...
PossibleEoarcheanRecordsoftheGeomagneticFieldPreservedintheIsuaSupracrustalBe...PossibleEoarcheanRecordsoftheGeomagneticFieldPreservedintheIsuaSupracrustalBe...
PossibleEoarcheanRecordsoftheGeomagneticFieldPreservedintheIsuaSupracrustalBe...
 
A relative description on Sonoporation.pdf
A relative description on Sonoporation.pdfA relative description on Sonoporation.pdf
A relative description on Sonoporation.pdf
 
Hubble Asteroid Hunter III. Physical properties of newly found asteroids
Hubble Asteroid Hunter III. Physical properties of newly found asteroidsHubble Asteroid Hunter III. Physical properties of newly found asteroids
Hubble Asteroid Hunter III. Physical properties of newly found asteroids
 
Artificial Intelligence In Microbiology by Dr. Prince C P
Artificial Intelligence In Microbiology by Dr. Prince C PArtificial Intelligence In Microbiology by Dr. Prince C P
Artificial Intelligence In Microbiology by Dr. Prince C P
 
Disentangling the origin of chemical differences using GHOST
Disentangling the origin of chemical differences using GHOSTDisentangling the origin of chemical differences using GHOST
Disentangling the origin of chemical differences using GHOST
 
Work, Energy and Power for class 10 ICSE Physics
Work, Energy and Power for class 10 ICSE PhysicsWork, Energy and Power for class 10 ICSE Physics
Work, Energy and Power for class 10 ICSE Physics
 
Luciferase in rDNA technology (biotechnology).pptx
Luciferase in rDNA technology (biotechnology).pptxLuciferase in rDNA technology (biotechnology).pptx
Luciferase in rDNA technology (biotechnology).pptx
 
Is RISC-V ready for HPC workload? Maybe?
Is RISC-V ready for HPC workload? Maybe?Is RISC-V ready for HPC workload? Maybe?
Is RISC-V ready for HPC workload? Maybe?
 
Behavioral Disorder: Schizophrenia & it's Case Study.pdf
Behavioral Disorder: Schizophrenia & it's Case Study.pdfBehavioral Disorder: Schizophrenia & it's Case Study.pdf
Behavioral Disorder: Schizophrenia & it's Case Study.pdf
 
Biopesticide (2).pptx .This slides helps to know the different types of biop...
Biopesticide (2).pptx  .This slides helps to know the different types of biop...Biopesticide (2).pptx  .This slides helps to know the different types of biop...
Biopesticide (2).pptx .This slides helps to know the different types of biop...
 
Ahmedabad Call Girls Service 9537192988 can satisfy every one of your dreams
Ahmedabad Call Girls Service 9537192988 can satisfy every one of your dreamsAhmedabad Call Girls Service 9537192988 can satisfy every one of your dreams
Ahmedabad Call Girls Service 9537192988 can satisfy every one of your dreams
 
Boyles law module in the grade 10 science
Boyles law module in the grade 10 scienceBoyles law module in the grade 10 science
Boyles law module in the grade 10 science
 
Natural Polymer Based Nanomaterials
Natural Polymer Based NanomaterialsNatural Polymer Based Nanomaterials
Natural Polymer Based Nanomaterials
 
Animal Communication- Auditory and Visual.pptx
Animal Communication- Auditory and Visual.pptxAnimal Communication- Auditory and Visual.pptx
Animal Communication- Auditory and Visual.pptx
 
Dashanga agada a formulation of Agada tantra dealt in 3 Rd year bams agada tanta
Dashanga agada a formulation of Agada tantra dealt in 3 Rd year bams agada tantaDashanga agada a formulation of Agada tantra dealt in 3 Rd year bams agada tanta
Dashanga agada a formulation of Agada tantra dealt in 3 Rd year bams agada tanta
 
SOLUBLE PATTERN RECOGNITION RECEPTORS.pptx
SOLUBLE PATTERN RECOGNITION RECEPTORS.pptxSOLUBLE PATTERN RECOGNITION RECEPTORS.pptx
SOLUBLE PATTERN RECOGNITION RECEPTORS.pptx
 

Interpretable Representation Learning with InfoGAN

  • 1. InfoGAN Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets
  • 2. Introduction  In ordinary GANs, the latent vector z has an arbitrary representation. The vector has no intrinsic meaning.  It would be desirable to make a more meaningful representation of the outputs.
  • 3. Disentangled Representation  We wish to disentangle the representation of the output images in the input latent vectors.  That is, we wish to make the values of the latent vector c correspond to features in the generated images.
  • 4. Information Theory A brief introduction to relevant concepts in information theory.
  • 5. Entropy Entropy can be intuitively understood the amount of information that a variable contains. It is borrowed from statistical thermodynamics and is directly analogous to the physical definition of randomness of particle states. 𝐻 𝑋 = − 𝑖=1 𝑛 𝑃 𝑥𝑖 log 𝑏 𝑃 𝑥𝑖 𝐻 𝑋 𝑌 = − 𝑖,𝑗 𝑝 𝑥𝑖, 𝑦𝑗 𝑙𝑜𝑔 𝑝(𝑥𝑖, 𝑦𝑖) 𝑝 𝑦𝑖
  • 6. Mutual Information: Definition  In information theory, I(X;Y), the mutual information between X and Y, measures the “amount of information” learned from knowledge of random variable Y about the random variable X.  The mutual information can be expressed as the difference of two entropy terms: I(X;Y) = H(X)-H(X|Y) = H(Y)-H(Y|X)
  • 7. Mutual Information: Interpretation  If X and Y are independent, then I(X;Y) = 0, because knowing one variable reveals nothing about the other.  If X and Y are related by a deterministic, invertible function, I(X;Y) is at its maximum.  I(X;Y) is the reduction of uncertainty in X when Y is observed and vice versa.
  • 8. Mutual Information: Implications  Formulation of a cost function using mutual information.  Information regularized mini-max game. min 𝐺 max 𝐷 𝑉𝐼 𝐷, 𝐺 = 𝑉 𝐷, 𝐺 − 𝜆𝐼(𝑐; 𝐺 𝑧, 𝑐 ) z: vector for representing intrinsic noise c: Latent code representing meaningful information.
  • 9. Implementation Variational Mutual Information Maximization and Approximations.
  • 10. Variational Mutual Information Maximization  In practice, 𝐼(𝑐; 𝐺 𝑧, 𝑐 ) cannot be maximized directly as this requires 𝑃(𝑐|𝑥), the posterior distribution.  We can obtain a lower bound for 𝐼(𝑐; 𝐺 𝑧, 𝑐 ) by using an auxiliary distribution, 𝑄 𝑐 𝑥 , to approximate 𝑃(𝑐|𝑥).  This technique is known as Variational Mutual Information Maximization. The equations are in the next slide.
  • 11. Variational Mutual Information Maximization  The Mutual Information is decomposed into its components.  Q is introduced to use the definition of KL divergence.  KL Divergence is always greater than 0 by definition.
  • 12. Variational Mutual Information Maximization  Although 𝐻(𝑐) can also be optimized, it is set as a constant for simplicity. This is done by drawing c from a fixed distribution.  The equality in this part is proven in the appendix of the paper. It holds under most conditions.
  • 13. Variational Mutual Information Maximization  By the proofs discussed previously, the actual information regularization is as follows. min 𝐺,𝑄 max 𝐷 𝑉𝐼𝑛𝑓𝑜𝐺𝐴𝑁 𝐷, 𝐺, 𝑄 = 𝑉 𝐷, 𝐺 − 𝜆𝐿𝐼(𝐺, 𝑄) 𝐿𝐼: Information Lower Bound 𝜆: Weighting Hyperparameter
  • 14. Practical Implementation  In practice, we use a neural network to represent Q.  KL Divergence is minimized when Q converges to P.  Q is just D with an extra FC layer. It outputs an estimate of the c latent vector(s).  𝐿𝐼(𝐺, 𝑄) has been observed to converge faster than GAN objectives.  InfoGAN thus comes essentially for free with GAN.
  • 15. Overview of the Architecture
  • 16. Additional explanation  Cross Entropy Loss is used on the estimate of c and the actual c (Some implementations use MSE).  For discrete cases, the outputs are softmax activations. For continuous cases, they may be sigmoid, tanh, or linear.  The original is much more complicated and does not directly output the estimate of the latent vector but estimates of its distribution (mean and standard deviation).
  • 17. Analysis  The information lower bound 𝐿𝐼 does not increase without information regularization.  The maximum appears to be 2.3 for this case.
  • 18. Results: MNIST The discrete latent vector input for digits is highly correlated with the output number. It can be used to classify MNIST with a 5% error rate, even when InfoGAN is trained without any labels. The continuous latent vector has found the angle of the characters. We confirm that the latent code validity by extending the range beyond the original. The ordinary GAN has learned nothing. There is no control over which vector learns what.
  • 19. Results: Faces Multiple vectors with a range between -1 and 1 are used. The 4 latent vectors appear to have captured azimuth, elevation, lighting, and width. There is smooth interpolation within and even beyond the range. Moreover, the other details change to make a much more natural image. It is not a simple case where only the target feature changes while the other factors remain unnaturally constant.
  • 23. Conclusion  InfoGAN can learn and interpret salient features of data without any labels or supervision.  InfoGAN can discover salient latent factors of variation automatically and produce a latent vector containing that information.