SlideShare ist ein Scribd-Unternehmen logo
1 von 15
A Baseline For Detecting Misclassified and Out-of-Distribution
Examples In Neural Networks
PR-190
Kang, MinGuk
mingukkang1994@gmail.com
Sep., 01, 2019
https://arxiv.org/abs/1610.02136
Unprecedented Successes(Motivation)
Image Classification Real-World Applications
https://arxiv.org/abs/1905.11946
Unprecedented Successes(Motivation)
https://www.researchgate.net/figure/Collage-of-some-medical-imaging-applications-in-which-deep-learning-has-achieved_fig1_313857891
Why Deep Neural Networks tend to be overconfident?
① Softmax Probabilities are computed
with the fast-growing exponential function
But… No Experimental Analysis
https://arxiv.org/pdf/1706.04599.pdf
Expected Calibration Error(ECE)
① Depth ↑
② Filters ↑
③ Batch Normalization 有
④ Weight Decay ↓
It remains future work to understand why these
trends affect calibration while improving accuracy.
Contributions of this Paper
1. They show the prediction probability of incorrect and out-of-distribution examples tends to be lower
than the prediction probability for correct examples.
2. These prediction probabilities form our detection baseline, and we demonstrate its efficacy through
various computer vision, natural language processing, and automatic speech recognition tasks.
3. They contribute one method which outperforms the baseline on some (but not all) tasks.
4. the designation of standard tasks and evaluation metrics for assessing the automatic detection of errors
and out-of-distribution examples.
Evaluation Metrics
In-distribution Fish: 99
Out-of-distribution Fish: 1
Cheating Neural Network: 99% accuracy!
So, Accuracy is not appropriate metric for out-of-distribution detection.
Evaluation Metrics
① AUROC(Area Under Receiver Operating Characteristic Curve) ① AUPR(Area Under Precision Recall Curve)
FPR(False Positive Rate):
𝐹𝑃
𝐹𝑃+𝑇𝑁
TPR(True Positive Rate):
𝑇𝑃
𝑇𝑃+𝐹𝑁
interpreted as the probability that a positive example has a greater
detector score/value than a negative example (Fawcett, 2005).
AUROC is not ideal when the positive class and negative class have
greatly differing base rates
Precision:
𝑇𝑃
𝑇𝑃+𝐹𝑃
Recall:
𝑇𝑃
𝑇𝑃+𝐹𝑁
interpreted as the probability that a positive example has a greater
detector score/value than a negative example (Fawcett, 2005).
AUROC is not ideal when the positive class and negative class have
greatly differing base rates
Experiments(Misclassified?)
Confi: 0.81 0.91 0.84 0.91 0.85 0.75 0.90 0.88
Average: 0.86
Confi: 0.90 0.95 0.85 0.95 0.92 0.88 0.95 0.86
Predict: 7
Actual: 7
Predict: 8
Actual: 8
Predict: 8
Actual: 8
Predict: 8
Actual: 8
Predict: 5
Actual: 5
Predict: 7
Actual: 7
Predict: 9
Actual: 9
Predict: 6
Actual: 6
Average: 0.91
Experiments(Out of Distribution)
Wide(40-4)
Prediction
Train
CIFAR10 Dataset
Test
Select Maximum Softmax Probability and
Use it as out-of-distribution score
Experiments(NLP)
Same Phenomenon was discovered in the NLP! Sentiment Classification
Text Categorization
Automatic Speech Recognition
Experimental Results of Sentiment Classification
Improved Method
Abnormality Module
1. Train a normal classifier and append an auxiliary decoder
which reconstructs the input with in-distribution dataset.
2. Froze the blue layer.
3. Train red layers on clean and noised training examples.
Finally the sigmoid output of the red layers scores how normal the input is
Improved Method
Abnormality Module
Abnormality Module is useful to detect out-of-distribution samples!
Expected Calibration Error(ECE)
① Depth ↑
② Filters ↑
③ Batch Normalization 有
④ Weight Decay ↓
It remains future work to understand why these
trends affect calibration while improving accuracy.
On Calibration of Modern Neural Networks
(2017.06.14)
(2016.10.07)
A Simple Unified Framework for Detecting Out-of-
Distribution Samples And Adversarial Attacks
(2018.07.10)
Training Confidence-Calibrated Classifiers for detecting
Out-of-Distribution samples
(2017.11.26)
Train Generative Adversarial Networks to generate
Boundary Samples.
Class(k)
Probability
1/k
Deep Anomaly Detection with Outlier Exposure
(2018.12.11)
Utilize Realistic Outliers instead of boundary samples
Class(k)
Probability
1/k
In-distribution dataset Out-of-Distribution dataset
Thank You!

Weitere ähnliche Inhalte

Was ist angesagt?

初めてのグラフカット
初めてのグラフカット初めてのグラフカット
初めてのグラフカット
Tsubasa Hirakawa
 

Was ist angesagt? (20)

【DL輪読会】GPT-4Technical Report
【DL輪読会】GPT-4Technical Report【DL輪読会】GPT-4Technical Report
【DL輪読会】GPT-4Technical Report
 
SIGMOD 2019 参加報告
SIGMOD 2019 参加報告SIGMOD 2019 参加報告
SIGMOD 2019 参加報告
 
方策勾配型強化学習の基礎と応用
方策勾配型強化学習の基礎と応用方策勾配型強化学習の基礎と応用
方策勾配型強化学習の基礎と応用
 
Math in Machine Learning / PCA and SVD with Applications
Math in Machine Learning / PCA and SVD with ApplicationsMath in Machine Learning / PCA and SVD with Applications
Math in Machine Learning / PCA and SVD with Applications
 
論文紹介-Multi-Objective Deep Reinforcement Learning
論文紹介-Multi-Objective Deep Reinforcement Learning論文紹介-Multi-Objective Deep Reinforcement Learning
論文紹介-Multi-Objective Deep Reinforcement Learning
 
金融時系列のための深層t過程回帰モデル
金融時系列のための深層t過程回帰モデル金融時系列のための深層t過程回帰モデル
金融時系列のための深層t過程回帰モデル
 
ようやく分かった!最尤推定とベイズ推定
ようやく分かった!最尤推定とベイズ推定ようやく分かった!最尤推定とベイズ推定
ようやく分かった!最尤推定とベイズ推定
 
学習時に使ってはいないデータの混入「リーケージを避ける」
学習時に使ってはいないデータの混入「リーケージを避ける」学習時に使ってはいないデータの混入「リーケージを避ける」
学習時に使ってはいないデータの混入「リーケージを避ける」
 
統計的因果推論への招待 -因果構造探索を中心に-
統計的因果推論への招待 -因果構造探索を中心に-統計的因果推論への招待 -因果構造探索を中心に-
統計的因果推論への招待 -因果構造探索を中心に-
 
畳み込みネットワークによる高次元信号復元と異分野融合への展開
畳み込みネットワークによる高次元信号復元と異分野融合への展開 畳み込みネットワークによる高次元信号復元と異分野融合への展開
畳み込みネットワークによる高次元信号復元と異分野融合への展開
 
音声認識と深層学習
音声認識と深層学習音声認識と深層学習
音声認識と深層学習
 
PyMCがあれば,ベイズ推定でもう泣いたりなんかしない
PyMCがあれば,ベイズ推定でもう泣いたりなんかしないPyMCがあれば,ベイズ推定でもう泣いたりなんかしない
PyMCがあれば,ベイズ推定でもう泣いたりなんかしない
 
はじめよう多変量解析~主成分分析編~
はじめよう多変量解析~主成分分析編~はじめよう多変量解析~主成分分析編~
はじめよう多変量解析~主成分分析編~
 
[DLHacks]Comet ML -機械学習のためのGitHub-
[DLHacks]Comet ML -機械学習のためのGitHub-[DLHacks]Comet ML -機械学習のためのGitHub-
[DLHacks]Comet ML -機械学習のためのGitHub-
 
初めてのグラフカット
初めてのグラフカット初めてのグラフカット
初めてのグラフカット
 
強化学習技術とゲーム AI 〜 今できる事と今後できて欲しい事 〜
強化学習技術とゲーム AI  〜 今できる事と今後できて欲しい事 〜強化学習技術とゲーム AI  〜 今できる事と今後できて欲しい事 〜
強化学習技術とゲーム AI 〜 今できる事と今後できて欲しい事 〜
 
機械学習におけるオンライン確率的最適化の理論
機械学習におけるオンライン確率的最適化の理論機械学習におけるオンライン確率的最適化の理論
機械学習におけるオンライン確率的最適化の理論
 
論文紹介「A Perspective View and Survey of Meta-Learning」
論文紹介「A Perspective View and Survey of Meta-Learning」論文紹介「A Perspective View and Survey of Meta-Learning」
論文紹介「A Perspective View and Survey of Meta-Learning」
 
[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
 
全力解説!Transformer
全力解説!Transformer全力解説!Transformer
全力解説!Transformer
 

Ähnlich wie PR-190: A Baseline For Detecting Misclassified and Out-of-Distribution Examples In Neural Networks

Annotated Bibliography .Guidelines Annotated Bibliograph.docx
Annotated Bibliography  .Guidelines Annotated Bibliograph.docxAnnotated Bibliography  .Guidelines Annotated Bibliograph.docx
Annotated Bibliography .Guidelines Annotated Bibliograph.docx
justine1simpson78276
 
Lecture7 Ml Machines That Can Learn
Lecture7 Ml Machines That Can LearnLecture7 Ml Machines That Can Learn
Lecture7 Ml Machines That Can Learn
Kodok Ngorex
 
BioAssay Express: Creating and exploiting assay metadata
BioAssay Express: Creating and exploiting assay metadataBioAssay Express: Creating and exploiting assay metadata
BioAssay Express: Creating and exploiting assay metadata
Philip Cheung
 
A PARTICLE SWARM OPTIMIZATION TECHNIQUE FOR GENERATING PAIRWISE TEST CASES
A PARTICLE SWARM OPTIMIZATION TECHNIQUE FOR GENERATING PAIRWISE TEST CASESA PARTICLE SWARM OPTIMIZATION TECHNIQUE FOR GENERATING PAIRWISE TEST CASES
A PARTICLE SWARM OPTIMIZATION TECHNIQUE FOR GENERATING PAIRWISE TEST CASES
Kula Sekhar Reddy Yerraguntla
 
Anomaly Detection Via PCA
Anomaly Detection Via PCAAnomaly Detection Via PCA
Anomaly Detection Via PCA
Deepak Kumar
 
A beginners guide to testing
A beginners guide to testingA beginners guide to testing
A beginners guide to testing
Philip Johnson
 

Ähnlich wie PR-190: A Baseline For Detecting Misclassified and Out-of-Distribution Examples In Neural Networks (20)

Annotated Bibliography .Guidelines Annotated Bibliograph.docx
Annotated Bibliography  .Guidelines Annotated Bibliograph.docxAnnotated Bibliography  .Guidelines Annotated Bibliograph.docx
Annotated Bibliography .Guidelines Annotated Bibliograph.docx
 
Machine Learning Interview Questions
Machine Learning Interview QuestionsMachine Learning Interview Questions
Machine Learning Interview Questions
 
Lecture7 Ml Machines That Can Learn
Lecture7 Ml Machines That Can LearnLecture7 Ml Machines That Can Learn
Lecture7 Ml Machines That Can Learn
 
Using the Machine to predict Testability
Using the Machine to predict TestabilityUsing the Machine to predict Testability
Using the Machine to predict Testability
 
Muwanika rogers (software testing) muni university
Muwanika rogers (software testing) muni universityMuwanika rogers (software testing) muni university
Muwanika rogers (software testing) muni university
 
Artificial Intelligence in Neurology.pptx
Artificial Intelligence in Neurology.pptxArtificial Intelligence in Neurology.pptx
Artificial Intelligence in Neurology.pptx
 
Barga Data Science lecture 9
Barga Data Science lecture 9Barga Data Science lecture 9
Barga Data Science lecture 9
 
BioAssay Express: Creating and exploiting assay metadata
BioAssay Express: Creating and exploiting assay metadataBioAssay Express: Creating and exploiting assay metadata
BioAssay Express: Creating and exploiting assay metadata
 
Neural networks, naïve bayes and decision tree machine learning
Neural networks, naïve bayes and decision tree machine learningNeural networks, naïve bayes and decision tree machine learning
Neural networks, naïve bayes and decision tree machine learning
 
A PARTICLE SWARM OPTIMIZATION TECHNIQUE FOR GENERATING PAIRWISE TEST CASES
A PARTICLE SWARM OPTIMIZATION TECHNIQUE FOR GENERATING PAIRWISE TEST CASESA PARTICLE SWARM OPTIMIZATION TECHNIQUE FOR GENERATING PAIRWISE TEST CASES
A PARTICLE SWARM OPTIMIZATION TECHNIQUE FOR GENERATING PAIRWISE TEST CASES
 
Barga Data Science lecture 10
Barga Data Science lecture 10Barga Data Science lecture 10
Barga Data Science lecture 10
 
Anomaly Detection Via PCA
Anomaly Detection Via PCAAnomaly Detection Via PCA
Anomaly Detection Via PCA
 
Credit Card Fraud Detection - Anomaly Detection
Credit Card Fraud Detection - Anomaly DetectionCredit Card Fraud Detection - Anomaly Detection
Credit Card Fraud Detection - Anomaly Detection
 
Key Insights Of Using Deep Learning To Analyze Healthcare Data | Workshop Fro...
Key Insights Of Using Deep Learning To Analyze Healthcare Data | Workshop Fro...Key Insights Of Using Deep Learning To Analyze Healthcare Data | Workshop Fro...
Key Insights Of Using Deep Learning To Analyze Healthcare Data | Workshop Fro...
 
Quality of data
Quality of dataQuality of data
Quality of data
 
Keerthi report
Keerthi reportKeerthi report
Keerthi report
 
A beginners guide to testing
A beginners guide to testingA beginners guide to testing
A beginners guide to testing
 
Model validation strategies ftc 2018
Model validation strategies ftc 2018Model validation strategies ftc 2018
Model validation strategies ftc 2018
 
Machine learning - session 4
Machine learning - session 4Machine learning - session 4
Machine learning - session 4
 
2015 osu-metagenome
2015 osu-metagenome2015 osu-metagenome
2015 osu-metagenome
 

Mehr von 강민국 강민국

Mehr von 강민국 강민국 (12)

Deeppermnet
DeeppermnetDeeppermnet
Deeppermnet
 
[Pr12] deep anomaly detection using geometric transformations
[Pr12] deep anomaly detection using geometric transformations[Pr12] deep anomaly detection using geometric transformations
[Pr12] deep anomaly detection using geometric transformations
 
[Pr12] self supervised gan
[Pr12] self supervised gan[Pr12] self supervised gan
[Pr12] self supervised gan
 
Ebgan
EbganEbgan
Ebgan
 
Anomaly detection
Anomaly detectionAnomaly detection
Anomaly detection
 
Deep Feature Consistent VAE
Deep Feature Consistent VAEDeep Feature Consistent VAE
Deep Feature Consistent VAE
 
[Probability for machine learning]
[Probability for machine learning][Probability for machine learning]
[Probability for machine learning]
 
Deep learning overview
Deep learning overviewDeep learning overview
Deep learning overview
 
Generative adversarial network
Generative adversarial networkGenerative adversarial network
Generative adversarial network
 
Variational AutoEncoder(VAE)
Variational AutoEncoder(VAE)Variational AutoEncoder(VAE)
Variational AutoEncoder(VAE)
 
Restricted boltzmann machine
Restricted boltzmann machineRestricted boltzmann machine
Restricted boltzmann machine
 
Backpropagation
BackpropagationBackpropagation
Backpropagation
 

Kürzlich hochgeladen

Abortion pills in Jeddah | +966572737505 | Get Cytotec
Abortion pills in Jeddah | +966572737505 | Get CytotecAbortion pills in Jeddah | +966572737505 | Get Cytotec
Abortion pills in Jeddah | +966572737505 | Get Cytotec
Abortion pills in Riyadh +966572737505 get cytotec
 
Top profile Call Girls In Indore [ 7014168258 ] Call Me For Genuine Models We...
Top profile Call Girls In Indore [ 7014168258 ] Call Me For Genuine Models We...Top profile Call Girls In Indore [ 7014168258 ] Call Me For Genuine Models We...
Top profile Call Girls In Indore [ 7014168258 ] Call Me For Genuine Models We...
gajnagarg
 
+97470301568>>weed for sale in qatar ,weed for sale in dubai,weed for sale in...
+97470301568>>weed for sale in qatar ,weed for sale in dubai,weed for sale in...+97470301568>>weed for sale in qatar ,weed for sale in dubai,weed for sale in...
+97470301568>>weed for sale in qatar ,weed for sale in dubai,weed for sale in...
Health
 
Top profile Call Girls In Vadodara [ 7014168258 ] Call Me For Genuine Models ...
Top profile Call Girls In Vadodara [ 7014168258 ] Call Me For Genuine Models ...Top profile Call Girls In Vadodara [ 7014168258 ] Call Me For Genuine Models ...
Top profile Call Girls In Vadodara [ 7014168258 ] Call Me For Genuine Models ...
gajnagarg
 
Top profile Call Girls In dimapur [ 7014168258 ] Call Me For Genuine Models W...
Top profile Call Girls In dimapur [ 7014168258 ] Call Me For Genuine Models W...Top profile Call Girls In dimapur [ 7014168258 ] Call Me For Genuine Models W...
Top profile Call Girls In dimapur [ 7014168258 ] Call Me For Genuine Models W...
gajnagarg
 
In Riyadh ((+919101817206)) Cytotec kit @ Abortion Pills Saudi Arabia
In Riyadh ((+919101817206)) Cytotec kit @ Abortion Pills Saudi ArabiaIn Riyadh ((+919101817206)) Cytotec kit @ Abortion Pills Saudi Arabia
In Riyadh ((+919101817206)) Cytotec kit @ Abortion Pills Saudi Arabia
ahmedjiabur940
 
Top profile Call Girls In bhavnagar [ 7014168258 ] Call Me For Genuine Models...
Top profile Call Girls In bhavnagar [ 7014168258 ] Call Me For Genuine Models...Top profile Call Girls In bhavnagar [ 7014168258 ] Call Me For Genuine Models...
Top profile Call Girls In bhavnagar [ 7014168258 ] Call Me For Genuine Models...
gajnagarg
 
Computer science Sql cheat sheet.pdf.pdf
Computer science Sql cheat sheet.pdf.pdfComputer science Sql cheat sheet.pdf.pdf
Computer science Sql cheat sheet.pdf.pdf
SayantanBiswas37
 
Top profile Call Girls In Latur [ 7014168258 ] Call Me For Genuine Models We ...
Top profile Call Girls In Latur [ 7014168258 ] Call Me For Genuine Models We ...Top profile Call Girls In Latur [ 7014168258 ] Call Me For Genuine Models We ...
Top profile Call Girls In Latur [ 7014168258 ] Call Me For Genuine Models We ...
gajnagarg
 
Reconciling Conflicting Data Curation Actions: Transparency Through Argument...
Reconciling Conflicting Data Curation Actions:  Transparency Through Argument...Reconciling Conflicting Data Curation Actions:  Transparency Through Argument...
Reconciling Conflicting Data Curation Actions: Transparency Through Argument...
Bertram Ludäscher
 
如何办理英国诺森比亚大学毕业证(NU毕业证书)成绩单原件一模一样
如何办理英国诺森比亚大学毕业证(NU毕业证书)成绩单原件一模一样如何办理英国诺森比亚大学毕业证(NU毕业证书)成绩单原件一模一样
如何办理英国诺森比亚大学毕业证(NU毕业证书)成绩单原件一模一样
wsppdmt
 

Kürzlich hochgeladen (20)

5CL-ADBA,5cladba, Chinese supplier, safety is guaranteed
5CL-ADBA,5cladba, Chinese supplier, safety is guaranteed5CL-ADBA,5cladba, Chinese supplier, safety is guaranteed
5CL-ADBA,5cladba, Chinese supplier, safety is guaranteed
 
Abortion pills in Jeddah | +966572737505 | Get Cytotec
Abortion pills in Jeddah | +966572737505 | Get CytotecAbortion pills in Jeddah | +966572737505 | Get Cytotec
Abortion pills in Jeddah | +966572737505 | Get Cytotec
 
Statistics notes ,it includes mean to index numbers
Statistics notes ,it includes mean to index numbersStatistics notes ,it includes mean to index numbers
Statistics notes ,it includes mean to index numbers
 
Top profile Call Girls In Indore [ 7014168258 ] Call Me For Genuine Models We...
Top profile Call Girls In Indore [ 7014168258 ] Call Me For Genuine Models We...Top profile Call Girls In Indore [ 7014168258 ] Call Me For Genuine Models We...
Top profile Call Girls In Indore [ 7014168258 ] Call Me For Genuine Models We...
 
+97470301568>>weed for sale in qatar ,weed for sale in dubai,weed for sale in...
+97470301568>>weed for sale in qatar ,weed for sale in dubai,weed for sale in...+97470301568>>weed for sale in qatar ,weed for sale in dubai,weed for sale in...
+97470301568>>weed for sale in qatar ,weed for sale in dubai,weed for sale in...
 
Top profile Call Girls In Vadodara [ 7014168258 ] Call Me For Genuine Models ...
Top profile Call Girls In Vadodara [ 7014168258 ] Call Me For Genuine Models ...Top profile Call Girls In Vadodara [ 7014168258 ] Call Me For Genuine Models ...
Top profile Call Girls In Vadodara [ 7014168258 ] Call Me For Genuine Models ...
 
Charbagh + Female Escorts Service in Lucknow | Starting ₹,5K To @25k with A/C...
Charbagh + Female Escorts Service in Lucknow | Starting ₹,5K To @25k with A/C...Charbagh + Female Escorts Service in Lucknow | Starting ₹,5K To @25k with A/C...
Charbagh + Female Escorts Service in Lucknow | Starting ₹,5K To @25k with A/C...
 
Top profile Call Girls In dimapur [ 7014168258 ] Call Me For Genuine Models W...
Top profile Call Girls In dimapur [ 7014168258 ] Call Me For Genuine Models W...Top profile Call Girls In dimapur [ 7014168258 ] Call Me For Genuine Models W...
Top profile Call Girls In dimapur [ 7014168258 ] Call Me For Genuine Models W...
 
In Riyadh ((+919101817206)) Cytotec kit @ Abortion Pills Saudi Arabia
In Riyadh ((+919101817206)) Cytotec kit @ Abortion Pills Saudi ArabiaIn Riyadh ((+919101817206)) Cytotec kit @ Abortion Pills Saudi Arabia
In Riyadh ((+919101817206)) Cytotec kit @ Abortion Pills Saudi Arabia
 
Top profile Call Girls In bhavnagar [ 7014168258 ] Call Me For Genuine Models...
Top profile Call Girls In bhavnagar [ 7014168258 ] Call Me For Genuine Models...Top profile Call Girls In bhavnagar [ 7014168258 ] Call Me For Genuine Models...
Top profile Call Girls In bhavnagar [ 7014168258 ] Call Me For Genuine Models...
 
20240412-SmartCityIndex-2024-Full-Report.pdf
20240412-SmartCityIndex-2024-Full-Report.pdf20240412-SmartCityIndex-2024-Full-Report.pdf
20240412-SmartCityIndex-2024-Full-Report.pdf
 
Computer science Sql cheat sheet.pdf.pdf
Computer science Sql cheat sheet.pdf.pdfComputer science Sql cheat sheet.pdf.pdf
Computer science Sql cheat sheet.pdf.pdf
 
Top profile Call Girls In Latur [ 7014168258 ] Call Me For Genuine Models We ...
Top profile Call Girls In Latur [ 7014168258 ] Call Me For Genuine Models We ...Top profile Call Girls In Latur [ 7014168258 ] Call Me For Genuine Models We ...
Top profile Call Girls In Latur [ 7014168258 ] Call Me For Genuine Models We ...
 
RESEARCH-FINAL-DEFENSE-PPT-TEMPLATE.pptx
RESEARCH-FINAL-DEFENSE-PPT-TEMPLATE.pptxRESEARCH-FINAL-DEFENSE-PPT-TEMPLATE.pptx
RESEARCH-FINAL-DEFENSE-PPT-TEMPLATE.pptx
 
Gomti Nagar & best call girls in Lucknow | 9548273370 Independent Escorts & D...
Gomti Nagar & best call girls in Lucknow | 9548273370 Independent Escorts & D...Gomti Nagar & best call girls in Lucknow | 9548273370 Independent Escorts & D...
Gomti Nagar & best call girls in Lucknow | 9548273370 Independent Escorts & D...
 
Reconciling Conflicting Data Curation Actions: Transparency Through Argument...
Reconciling Conflicting Data Curation Actions:  Transparency Through Argument...Reconciling Conflicting Data Curation Actions:  Transparency Through Argument...
Reconciling Conflicting Data Curation Actions: Transparency Through Argument...
 
Predicting HDB Resale Prices - Conducting Linear Regression Analysis With Orange
Predicting HDB Resale Prices - Conducting Linear Regression Analysis With OrangePredicting HDB Resale Prices - Conducting Linear Regression Analysis With Orange
Predicting HDB Resale Prices - Conducting Linear Regression Analysis With Orange
 
TrafficWave Generator Will Instantly drive targeted and engaging traffic back...
TrafficWave Generator Will Instantly drive targeted and engaging traffic back...TrafficWave Generator Will Instantly drive targeted and engaging traffic back...
TrafficWave Generator Will Instantly drive targeted and engaging traffic back...
 
如何办理英国诺森比亚大学毕业证(NU毕业证书)成绩单原件一模一样
如何办理英国诺森比亚大学毕业证(NU毕业证书)成绩单原件一模一样如何办理英国诺森比亚大学毕业证(NU毕业证书)成绩单原件一模一样
如何办理英国诺森比亚大学毕业证(NU毕业证书)成绩单原件一模一样
 
Top Call Girls in Balaghat 9332606886Call Girls Advance Cash On Delivery Ser...
Top Call Girls in Balaghat  9332606886Call Girls Advance Cash On Delivery Ser...Top Call Girls in Balaghat  9332606886Call Girls Advance Cash On Delivery Ser...
Top Call Girls in Balaghat 9332606886Call Girls Advance Cash On Delivery Ser...
 

PR-190: A Baseline For Detecting Misclassified and Out-of-Distribution Examples In Neural Networks

  • 1. A Baseline For Detecting Misclassified and Out-of-Distribution Examples In Neural Networks PR-190 Kang, MinGuk mingukkang1994@gmail.com Sep., 01, 2019 https://arxiv.org/abs/1610.02136
  • 2. Unprecedented Successes(Motivation) Image Classification Real-World Applications https://arxiv.org/abs/1905.11946
  • 4. Why Deep Neural Networks tend to be overconfident? ① Softmax Probabilities are computed with the fast-growing exponential function But… No Experimental Analysis https://arxiv.org/pdf/1706.04599.pdf Expected Calibration Error(ECE) ① Depth ↑ ② Filters ↑ ③ Batch Normalization 有 ④ Weight Decay ↓ It remains future work to understand why these trends affect calibration while improving accuracy.
  • 5. Contributions of this Paper 1. They show the prediction probability of incorrect and out-of-distribution examples tends to be lower than the prediction probability for correct examples. 2. These prediction probabilities form our detection baseline, and we demonstrate its efficacy through various computer vision, natural language processing, and automatic speech recognition tasks. 3. They contribute one method which outperforms the baseline on some (but not all) tasks. 4. the designation of standard tasks and evaluation metrics for assessing the automatic detection of errors and out-of-distribution examples.
  • 6. Evaluation Metrics In-distribution Fish: 99 Out-of-distribution Fish: 1 Cheating Neural Network: 99% accuracy! So, Accuracy is not appropriate metric for out-of-distribution detection.
  • 7. Evaluation Metrics ① AUROC(Area Under Receiver Operating Characteristic Curve) ① AUPR(Area Under Precision Recall Curve) FPR(False Positive Rate): 𝐹𝑃 𝐹𝑃+𝑇𝑁 TPR(True Positive Rate): 𝑇𝑃 𝑇𝑃+𝐹𝑁 interpreted as the probability that a positive example has a greater detector score/value than a negative example (Fawcett, 2005). AUROC is not ideal when the positive class and negative class have greatly differing base rates Precision: 𝑇𝑃 𝑇𝑃+𝐹𝑃 Recall: 𝑇𝑃 𝑇𝑃+𝐹𝑁 interpreted as the probability that a positive example has a greater detector score/value than a negative example (Fawcett, 2005). AUROC is not ideal when the positive class and negative class have greatly differing base rates
  • 8. Experiments(Misclassified?) Confi: 0.81 0.91 0.84 0.91 0.85 0.75 0.90 0.88 Average: 0.86 Confi: 0.90 0.95 0.85 0.95 0.92 0.88 0.95 0.86 Predict: 7 Actual: 7 Predict: 8 Actual: 8 Predict: 8 Actual: 8 Predict: 8 Actual: 8 Predict: 5 Actual: 5 Predict: 7 Actual: 7 Predict: 9 Actual: 9 Predict: 6 Actual: 6 Average: 0.91
  • 9. Experiments(Out of Distribution) Wide(40-4) Prediction Train CIFAR10 Dataset Test Select Maximum Softmax Probability and Use it as out-of-distribution score
  • 10. Experiments(NLP) Same Phenomenon was discovered in the NLP! Sentiment Classification Text Categorization Automatic Speech Recognition Experimental Results of Sentiment Classification
  • 11. Improved Method Abnormality Module 1. Train a normal classifier and append an auxiliary decoder which reconstructs the input with in-distribution dataset. 2. Froze the blue layer. 3. Train red layers on clean and noised training examples. Finally the sigmoid output of the red layers scores how normal the input is
  • 12. Improved Method Abnormality Module Abnormality Module is useful to detect out-of-distribution samples!
  • 13. Expected Calibration Error(ECE) ① Depth ↑ ② Filters ↑ ③ Batch Normalization 有 ④ Weight Decay ↓ It remains future work to understand why these trends affect calibration while improving accuracy. On Calibration of Modern Neural Networks (2017.06.14) (2016.10.07)
  • 14. A Simple Unified Framework for Detecting Out-of- Distribution Samples And Adversarial Attacks (2018.07.10) Training Confidence-Calibrated Classifiers for detecting Out-of-Distribution samples (2017.11.26) Train Generative Adversarial Networks to generate Boundary Samples. Class(k) Probability 1/k
  • 15. Deep Anomaly Detection with Outlier Exposure (2018.12.11) Utilize Realistic Outliers instead of boundary samples Class(k) Probability 1/k In-distribution dataset Out-of-Distribution dataset Thank You!