SlideShare a Scribd company logo
1 of 15
Download to read offline
PU Learning for Aurora Image
Segmentation
https://github.com/RxstydnR/PUlearning_segmentation
PU Learning ( Positive Unlabeled learning )
4
theory 0
PN Learning PU Learning
!
"
!
!
PU Learning
5
theory 1
=
   
   
p(y = 1|x) =
p(s = 1|x)
p(s = 1|y = 1)
:
: Positive,
: Negative,
: Positive,
: Negative,
x
y = 1
y = 0
s = 1
s = 0
x
g(x) = p(s = 1 ∣ x) = p(y = 1 ∧ s = 1 ∣ x)
= p(y = 1 ∣ x) p(s = 1 ∣ y = 1,x)
= p(y = 1 ∣ x) p(s = 1 ∣ y = 1)
= f(x) c
PU Learning
6
theory 2
:
: Positive,
: Negative,
: Positive,
: Negative,
x
y = 1
y = 0
s = 1
s = 0
p(y = 1|x) =
p(s = 1|x)
p(s = 1|y = 1)
c = p(s = 1|y = 1)
f(x) =
g(x)
c
=
   
   
≤ 1
PU Learning
7
theory 3
w(x) = p(y = 1 ∣ x, s = 0)
= ⋯
=
1 − c
c
p(s = 1 ∣ x)
1 − p(s = 1 ∣ x)
:
: Positive,
: Negative,
: Positive,
: Negative,
x
y = 1
y = 0
s = 1
s = 0
p(y = 1|x) =
p(s = 1|x)
p(s = 1|y = 1)
=
g(x)
c
c = p(s = 1|y = 1)
=
1
n ∑
x∈P
g(x) =
1
n ∑
x∈P
p(s = 1 ∣ x)
PU Learning
8
theory 4-1
1. , .
2. hold-out .
3. , .
4. .
(X, s) s g(x)
g(x) c =
∑
x∈P
p(s = 1 ∣ x)
y
f(x) = p(y = 1|x) =
p(s = 1|x)
p(s = 1|y = 1)
=
g(x)
c
PU Learning
9
theory 4-2
1. , .
2. hold-out .
3. , .
4. .
(X, s) s g(x)
g(x) c =
∑
x∈P
p(s = 1 ∣ x)
y
f(x) = p(y = 1|x) =
p(s = 1|x)
p(s = 1|y = 1)
=
g(x)
c
PU Learning
10
theory 4-3
1. , .
2. hold-out .
3. , .
4. .
(X, s) s g(x)
g(x) c =
∑
x∈P
p(s = 1 ∣ x)
y
w(x) = p(y = 1 ∣ x, s = 0) =
1 − c
c
p(s = 1 ∣ x)
1 − p(s = 1 ∣ x)
PU Learning
11
theory 4-4
1. , .
2. hold-out .
3. , .
4. .
(X, s) s g(x)
g(x) c =
∑
x∈P
p(s = 1 ∣ x)
y
PU Learning
12
theory 4-5
1. , .
2. hold-out .
3. , .
4. .
(X, s) s g(x)
g(x) c =
∑
x∈P
p(s = 1 ∣ x)
y
No Weighted PU Learning
Weighted PU Learning
w(x) = p(y = 1 ∣ x, s = 0) =
1 − c
c
p(s = 1 ∣ x)
1 − p(s = 1 ∣ x)
f(x) = p(y = 1|x) =
p(s = 1|x)
p(s = 1|y = 1)
=
g(x)
c
,
CV
CV
CV
PU Learning - Artificial blob data
13
Example
Data: 6000
(P: 3000, N: 3000)
Data: 6000
(P: 300, U: 5700)
PU Learning - Artificial blob data
14
Example
RandomForest PU Learning + RandomForest
PU Learning - Breast cancer dataset
15
Example
Positive
Train: 455 (P:159, N: 296)
Test: 228 (P:80, N: 148)
Data: 683 (P:239, N: 444)
P : 39
U : 416
PU Learning for DeepLearning
16
theory 5
:
: Positive
: Negative
:
: ,
X
πp = p(Y = + 1)
πn = p(Y = − 1)
g
ℓ(t, y) t y
Risk Estimator PU Learning
̂Rpu(g) = πp
̂R+
p (g) − πp
̂R−
p (g) + ̂R−
u (g)
R−
u (g) = 𝔼X∼p(x)[ℓ(g(X), − 1)]
R+
p (g) = 𝔼p[ℓ(g(X), + 1)],  where 𝔼p[ ⋅ ] = 𝔼X∼pp
[ ⋅ ]
R−
p (g) = 𝔼p[ℓ(g(X), − 1)],  where 𝔼p[ ⋅ ] = 𝔼X∼pp
[ ⋅ ]
∑
θ∈Θ
R(θ, δ)π(θ)
,
PU Learning for DeepLearning
17
theory 6
Risk Estimator PN Learning
̂Rpn(g) = πp
̂R+
p (g) + πn
̂R−
n (g)
R−
n (g) = 𝔼n[ℓ(g(X), − 1)],  where 𝔼n[ ⋅ ] = 𝔼X∼pn
[ ⋅ ]
R+
p (g) = 𝔼p[ℓ(g(X), + 1)],  where 𝔼p[ ⋅ ] = 𝔼X∼pp
[ ⋅ ]
Risk Estimator PU Learning
̂Rpu(g) = πp
̂R+
p (g) − πp
̂R−
p (g) + ̂R−
u (g)
R−
u (g) = 𝔼X∼p(x)[ℓ(g(X), − 1)]
R+
p (g) = 𝔼p[ℓ(g(X), + 1)],  where 𝔼p[ ⋅ ] = 𝔼X∼pp
[ ⋅ ]
R−
p (g) = 𝔼p[ℓ(g(X), − 1)],  where 𝔼p[ ⋅ ] = 𝔼X∼pp
[ ⋅ ]
{
πnpn(x) = p(x) − πppp(x)
πnR−
n (g) = R−
u (g) − πpR−
p (g)
,

More Related Content

What's hot

深層生成モデルを用いたマルチモーダルデータの半教師あり学習
深層生成モデルを用いたマルチモーダルデータの半教師あり学習深層生成モデルを用いたマルチモーダルデータの半教師あり学習
深層生成モデルを用いたマルチモーダルデータの半教師あり学習
Masahiro Suzuki
 
HypergroupsAssociationSchemes_leejuntaek
HypergroupsAssociationSchemes_leejuntaekHypergroupsAssociationSchemes_leejuntaek
HypergroupsAssociationSchemes_leejuntaek
Jun Taek Lee
 

What's hot (20)

Side 2019 #7
Side 2019 #7Side 2019 #7
Side 2019 #7
 
Truth, deduction, computation lecture e
Truth, deduction, computation   lecture eTruth, deduction, computation   lecture e
Truth, deduction, computation lecture e
 
Cii integral 5
Cii integral 5Cii integral 5
Cii integral 5
 
Slides simplexe
Slides simplexeSlides simplexe
Slides simplexe
 
深層生成モデルを用いたマルチモーダルデータの半教師あり学習
深層生成モデルを用いたマルチモーダルデータの半教師あり学習深層生成モデルを用いたマルチモーダルデータの半教師あり学習
深層生成モデルを用いたマルチモーダルデータの半教師あり学習
 
[DL輪読会]GANとエネルギーベースモデル
[DL輪読会]GANとエネルギーベースモデル[DL輪読会]GANとエネルギーベースモデル
[DL輪読会]GANとエネルギーベースモデル
 
Hands-On Algorithms for Predictive Modeling
Hands-On Algorithms for Predictive ModelingHands-On Algorithms for Predictive Modeling
Hands-On Algorithms for Predictive Modeling
 
【DL輪読会】SUMO: Unbiased Estimation of Log Marginal Probability for Latent Varia...
【DL輪読会】SUMO: Unbiased Estimation of Log Marginal Probability for Latent Varia...【DL輪読会】SUMO: Unbiased Estimation of Log Marginal Probability for Latent Varia...
【DL輪読会】SUMO: Unbiased Estimation of Log Marginal Probability for Latent Varia...
 
Slides econometrics-2018-graduate-3
Slides econometrics-2018-graduate-3Slides econometrics-2018-graduate-3
Slides econometrics-2018-graduate-3
 
【DL輪読会】NeRF-VAE: A Geometry Aware 3D Scene Generative Model
【DL輪読会】NeRF-VAE: A Geometry Aware 3D Scene Generative Model【DL輪読会】NeRF-VAE: A Geometry Aware 3D Scene Generative Model
【DL輪読会】NeRF-VAE: A Geometry Aware 3D Scene Generative Model
 
HypergroupsAssociationSchemes_leejuntaek
HypergroupsAssociationSchemes_leejuntaekHypergroupsAssociationSchemes_leejuntaek
HypergroupsAssociationSchemes_leejuntaek
 
Slides lln-risques
Slides lln-risquesSlides lln-risques
Slides lln-risques
 
Side 2019, part 2
Side 2019, part 2Side 2019, part 2
Side 2019, part 2
 
Sildes buenos aires
Sildes buenos airesSildes buenos aires
Sildes buenos aires
 
Side 2019, part 1
Side 2019, part 1Side 2019, part 1
Side 2019, part 1
 
Slides econ-lm
Slides econ-lmSlides econ-lm
Slides econ-lm
 
Approximate Matching (String Algorithms 2007)
Approximate Matching (String Algorithms 2007)Approximate Matching (String Algorithms 2007)
Approximate Matching (String Algorithms 2007)
 
Teaching Population Genetics with R
Teaching Population Genetics with RTeaching Population Genetics with R
Teaching Population Genetics with R
 
Slides barcelona Machine Learning
Slides barcelona Machine LearningSlides barcelona Machine Learning
Slides barcelona Machine Learning
 
Side 2019 #10
Side 2019 #10Side 2019 #10
Side 2019 #10
 

Similar to PU Learning

ゲーム理論BASIC 第23回 -ベイジアンゲームにおける戦略と均衡-
ゲーム理論BASIC 第23回 -ベイジアンゲームにおける戦略と均衡-ゲーム理論BASIC 第23回 -ベイジアンゲームにおける戦略と均衡-
ゲーム理論BASIC 第23回 -ベイジアンゲームにおける戦略と均衡-
ssusere0a682
 

Similar to PU Learning (20)

Orthogonal basis and gram schmidth process
Orthogonal basis and gram schmidth processOrthogonal basis and gram schmidth process
Orthogonal basis and gram schmidth process
 
「ベータ分布の謎に迫る」第6回 プログラマのための数学勉強会 LT資料
「ベータ分布の謎に迫る」第6回 プログラマのための数学勉強会 LT資料「ベータ分布の謎に迫る」第6回 プログラマのための数学勉強会 LT資料
「ベータ分布の謎に迫る」第6回 プログラマのための数学勉強会 LT資料
 
On Certain Classess of Multivalent Functions
On Certain Classess of Multivalent Functions On Certain Classess of Multivalent Functions
On Certain Classess of Multivalent Functions
 
cps170_bayes_nets.ppt
cps170_bayes_nets.pptcps170_bayes_nets.ppt
cps170_bayes_nets.ppt
 
Madrid easy
Madrid easyMadrid easy
Madrid easy
 
Vancouver18
Vancouver18Vancouver18
Vancouver18
 
Classification
ClassificationClassification
Classification
 
Slides univ-van-amsterdam
Slides univ-van-amsterdamSlides univ-van-amsterdam
Slides univ-van-amsterdam
 
sada_pres
sada_pressada_pres
sada_pres
 
Slides ACTINFO 2016
Slides ACTINFO 2016Slides ACTINFO 2016
Slides ACTINFO 2016
 
Probability QA 7
Probability QA 7Probability QA 7
Probability QA 7
 
Functions of severable variables
Functions of severable variablesFunctions of severable variables
Functions of severable variables
 
QMC Program: Trends and Advances in Monte Carlo Sampling Algorithms Workshop,...
QMC Program: Trends and Advances in Monte Carlo Sampling Algorithms Workshop,...QMC Program: Trends and Advances in Monte Carlo Sampling Algorithms Workshop,...
QMC Program: Trends and Advances in Monte Carlo Sampling Algorithms Workshop,...
 
A/B Testing for Game Design
A/B Testing for Game DesignA/B Testing for Game Design
A/B Testing for Game Design
 
Optimization of probabilistic argumentation with Markov processes
Optimization of probabilistic argumentation with Markov processesOptimization of probabilistic argumentation with Markov processes
Optimization of probabilistic argumentation with Markov processes
 
Wu Mamber (String Algorithms 2007)
Wu  Mamber (String Algorithms 2007)Wu  Mamber (String Algorithms 2007)
Wu Mamber (String Algorithms 2007)
 
Hierarchical Reinforcement Learning with Option-Critic Architecture
Hierarchical Reinforcement Learning with Option-Critic ArchitectureHierarchical Reinforcement Learning with Option-Critic Architecture
Hierarchical Reinforcement Learning with Option-Critic Architecture
 
ゲーム理論BASIC 第23回 -ベイジアンゲームにおける戦略と均衡-
ゲーム理論BASIC 第23回 -ベイジアンゲームにおける戦略と均衡-ゲーム理論BASIC 第23回 -ベイジアンゲームにおける戦略と均衡-
ゲーム理論BASIC 第23回 -ベイジアンゲームにおける戦略と均衡-
 
talk MCMC & SMC 2004
talk MCMC & SMC 2004talk MCMC & SMC 2004
talk MCMC & SMC 2004
 
11.[29 35]a unique common fixed point theorem under psi varphi contractive co...
11.[29 35]a unique common fixed point theorem under psi varphi contractive co...11.[29 35]a unique common fixed point theorem under psi varphi contractive co...
11.[29 35]a unique common fixed point theorem under psi varphi contractive co...
 

Recently uploaded

Artificial Intelligence: Facts and Myths
Artificial Intelligence: Facts and MythsArtificial Intelligence: Facts and Myths
Artificial Intelligence: Facts and Myths
Joaquim Jorge
 
IAC 2024 - IA Fast Track to Search Focused AI Solutions
IAC 2024 - IA Fast Track to Search Focused AI SolutionsIAC 2024 - IA Fast Track to Search Focused AI Solutions
IAC 2024 - IA Fast Track to Search Focused AI Solutions
Enterprise Knowledge
 

Recently uploaded (20)

Finology Group – Insurtech Innovation Award 2024
Finology Group – Insurtech Innovation Award 2024Finology Group – Insurtech Innovation Award 2024
Finology Group – Insurtech Innovation Award 2024
 
Boost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivityBoost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivity
 
Scaling API-first – The story of a global engineering organization
Scaling API-first – The story of a global engineering organizationScaling API-first – The story of a global engineering organization
Scaling API-first – The story of a global engineering organization
 
Artificial Intelligence: Facts and Myths
Artificial Intelligence: Facts and MythsArtificial Intelligence: Facts and Myths
Artificial Intelligence: Facts and Myths
 
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
 
Automating Google Workspace (GWS) & more with Apps Script
Automating Google Workspace (GWS) & more with Apps ScriptAutomating Google Workspace (GWS) & more with Apps Script
Automating Google Workspace (GWS) & more with Apps Script
 
Understanding Discord NSFW Servers A Guide for Responsible Users.pdf
Understanding Discord NSFW Servers A Guide for Responsible Users.pdfUnderstanding Discord NSFW Servers A Guide for Responsible Users.pdf
Understanding Discord NSFW Servers A Guide for Responsible Users.pdf
 
presentation ICT roal in 21st century education
presentation ICT roal in 21st century educationpresentation ICT roal in 21st century education
presentation ICT roal in 21st century education
 
TrustArc Webinar - Stay Ahead of US State Data Privacy Law Developments
TrustArc Webinar - Stay Ahead of US State Data Privacy Law DevelopmentsTrustArc Webinar - Stay Ahead of US State Data Privacy Law Developments
TrustArc Webinar - Stay Ahead of US State Data Privacy Law Developments
 
Driving Behavioral Change for Information Management through Data-Driven Gree...
Driving Behavioral Change for Information Management through Data-Driven Gree...Driving Behavioral Change for Information Management through Data-Driven Gree...
Driving Behavioral Change for Information Management through Data-Driven Gree...
 
Presentation on how to chat with PDF using ChatGPT code interpreter
Presentation on how to chat with PDF using ChatGPT code interpreterPresentation on how to chat with PDF using ChatGPT code interpreter
Presentation on how to chat with PDF using ChatGPT code interpreter
 
08448380779 Call Girls In Civil Lines Women Seeking Men
08448380779 Call Girls In Civil Lines Women Seeking Men08448380779 Call Girls In Civil Lines Women Seeking Men
08448380779 Call Girls In Civil Lines Women Seeking Men
 
IAC 2024 - IA Fast Track to Search Focused AI Solutions
IAC 2024 - IA Fast Track to Search Focused AI SolutionsIAC 2024 - IA Fast Track to Search Focused AI Solutions
IAC 2024 - IA Fast Track to Search Focused AI Solutions
 
How to Troubleshoot Apps for the Modern Connected Worker
How to Troubleshoot Apps for the Modern Connected WorkerHow to Troubleshoot Apps for the Modern Connected Worker
How to Troubleshoot Apps for the Modern Connected Worker
 
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
 
A Domino Admins Adventures (Engage 2024)
A Domino Admins Adventures (Engage 2024)A Domino Admins Adventures (Engage 2024)
A Domino Admins Adventures (Engage 2024)
 
Strategies for Landing an Oracle DBA Job as a Fresher
Strategies for Landing an Oracle DBA Job as a FresherStrategies for Landing an Oracle DBA Job as a Fresher
Strategies for Landing an Oracle DBA Job as a Fresher
 
Handwritten Text Recognition for manuscripts and early printed texts
Handwritten Text Recognition for manuscripts and early printed textsHandwritten Text Recognition for manuscripts and early printed texts
Handwritten Text Recognition for manuscripts and early printed texts
 
[2024]Digital Global Overview Report 2024 Meltwater.pdf
[2024]Digital Global Overview Report 2024 Meltwater.pdf[2024]Digital Global Overview Report 2024 Meltwater.pdf
[2024]Digital Global Overview Report 2024 Meltwater.pdf
 
08448380779 Call Girls In Diplomatic Enclave Women Seeking Men
08448380779 Call Girls In Diplomatic Enclave Women Seeking Men08448380779 Call Girls In Diplomatic Enclave Women Seeking Men
08448380779 Call Girls In Diplomatic Enclave Women Seeking Men
 

PU Learning

  • 1. PU Learning for Aurora Image Segmentation https://github.com/RxstydnR/PUlearning_segmentation
  • 2. PU Learning ( Positive Unlabeled learning ) 4 theory 0 PN Learning PU Learning ! " ! !
  • 3. PU Learning 5 theory 1 =         p(y = 1|x) = p(s = 1|x) p(s = 1|y = 1) : : Positive, : Negative, : Positive, : Negative, x y = 1 y = 0 s = 1 s = 0
  • 4. x g(x) = p(s = 1 ∣ x) = p(y = 1 ∧ s = 1 ∣ x) = p(y = 1 ∣ x) p(s = 1 ∣ y = 1,x) = p(y = 1 ∣ x) p(s = 1 ∣ y = 1) = f(x) c PU Learning 6 theory 2 : : Positive, : Negative, : Positive, : Negative, x y = 1 y = 0 s = 1 s = 0 p(y = 1|x) = p(s = 1|x) p(s = 1|y = 1) c = p(s = 1|y = 1) f(x) = g(x) c =         ≤ 1
  • 5. PU Learning 7 theory 3 w(x) = p(y = 1 ∣ x, s = 0) = ⋯ = 1 − c c p(s = 1 ∣ x) 1 − p(s = 1 ∣ x) : : Positive, : Negative, : Positive, : Negative, x y = 1 y = 0 s = 1 s = 0 p(y = 1|x) = p(s = 1|x) p(s = 1|y = 1) = g(x) c c = p(s = 1|y = 1) = 1 n ∑ x∈P g(x) = 1 n ∑ x∈P p(s = 1 ∣ x)
  • 6. PU Learning 8 theory 4-1 1. , . 2. hold-out . 3. , . 4. . (X, s) s g(x) g(x) c = ∑ x∈P p(s = 1 ∣ x) y f(x) = p(y = 1|x) = p(s = 1|x) p(s = 1|y = 1) = g(x) c
  • 7. PU Learning 9 theory 4-2 1. , . 2. hold-out . 3. , . 4. . (X, s) s g(x) g(x) c = ∑ x∈P p(s = 1 ∣ x) y f(x) = p(y = 1|x) = p(s = 1|x) p(s = 1|y = 1) = g(x) c
  • 8. PU Learning 10 theory 4-3 1. , . 2. hold-out . 3. , . 4. . (X, s) s g(x) g(x) c = ∑ x∈P p(s = 1 ∣ x) y w(x) = p(y = 1 ∣ x, s = 0) = 1 − c c p(s = 1 ∣ x) 1 − p(s = 1 ∣ x)
  • 9. PU Learning 11 theory 4-4 1. , . 2. hold-out . 3. , . 4. . (X, s) s g(x) g(x) c = ∑ x∈P p(s = 1 ∣ x) y
  • 10. PU Learning 12 theory 4-5 1. , . 2. hold-out . 3. , . 4. . (X, s) s g(x) g(x) c = ∑ x∈P p(s = 1 ∣ x) y No Weighted PU Learning Weighted PU Learning w(x) = p(y = 1 ∣ x, s = 0) = 1 − c c p(s = 1 ∣ x) 1 − p(s = 1 ∣ x) f(x) = p(y = 1|x) = p(s = 1|x) p(s = 1|y = 1) = g(x) c , CV CV CV
  • 11. PU Learning - Artificial blob data 13 Example Data: 6000 (P: 3000, N: 3000) Data: 6000 (P: 300, U: 5700)
  • 12. PU Learning - Artificial blob data 14 Example RandomForest PU Learning + RandomForest
  • 13. PU Learning - Breast cancer dataset 15 Example Positive Train: 455 (P:159, N: 296) Test: 228 (P:80, N: 148) Data: 683 (P:239, N: 444) P : 39 U : 416
  • 14. PU Learning for DeepLearning 16 theory 5 : : Positive : Negative : : , X πp = p(Y = + 1) πn = p(Y = − 1) g ℓ(t, y) t y Risk Estimator PU Learning ̂Rpu(g) = πp ̂R+ p (g) − πp ̂R− p (g) + ̂R− u (g) R− u (g) = 𝔼X∼p(x)[ℓ(g(X), − 1)] R+ p (g) = 𝔼p[ℓ(g(X), + 1)],  where 𝔼p[ ⋅ ] = 𝔼X∼pp [ ⋅ ] R− p (g) = 𝔼p[ℓ(g(X), − 1)],  where 𝔼p[ ⋅ ] = 𝔼X∼pp [ ⋅ ] ∑ θ∈Θ R(θ, δ)π(θ) ,
  • 15. PU Learning for DeepLearning 17 theory 6 Risk Estimator PN Learning ̂Rpn(g) = πp ̂R+ p (g) + πn ̂R− n (g) R− n (g) = 𝔼n[ℓ(g(X), − 1)],  where 𝔼n[ ⋅ ] = 𝔼X∼pn [ ⋅ ] R+ p (g) = 𝔼p[ℓ(g(X), + 1)],  where 𝔼p[ ⋅ ] = 𝔼X∼pp [ ⋅ ] Risk Estimator PU Learning ̂Rpu(g) = πp ̂R+ p (g) − πp ̂R− p (g) + ̂R− u (g) R− u (g) = 𝔼X∼p(x)[ℓ(g(X), − 1)] R+ p (g) = 𝔼p[ℓ(g(X), + 1)],  where 𝔼p[ ⋅ ] = 𝔼X∼pp [ ⋅ ] R− p (g) = 𝔼p[ℓ(g(X), − 1)],  where 𝔼p[ ⋅ ] = 𝔼X∼pp [ ⋅ ] { πnpn(x) = p(x) − πppp(x) πnR− n (g) = R− u (g) − πpR− p (g) ,