SlideShare a Scribd company logo
1 of 36
Download to read offline
Understanding black-box
predictions via influence
functions
XIE Ruiming
Outline
• Background
• Taylor's Formula
• Newton's Method
• Introduction
• Influence Function
• Definition
• Efficiently Calculating Influence
• Validation and Extensions
• Use cases of influence function
Background : Taylor
• Taylor's theorem gives an approximation of a k-times
differentiable function around a given point by a k-th
order Taylor polynomial
• Linear approximation
• Quadratic approximation
Background : Newton
• Find x:F(x) = 0 through iteration.
• Recall Taylor’s Formula
• F(a) ≈ F(Xn) + F’(Xn)(a – Xn)
• Set F(a) = 0, get a = Xn – F(Xn)/F’(Xn)
• Newton in optimizing
• X = argmin(F(x)), then F’(X) = 0
• Doing Newton’s method with F’(X)
• Xn+1 = Xn – F’(Xn)/F’’(Xn)
Background : Newton
• X = parameters
Introduction
Introduction
• Why did the model make this prediction?
Introduction
• Why did the model make this prediction?
• Retrieving images that maximally activate a neuron [Girshick et
al. 2014]
Introduction
• Why did the model make this prediction?
• Retrieving images that maximally activate a neuron [Girshick et
al. 2014]
• Finding the most influential part from the image [Zhou et al. 2016]
Introduction
• Why did the model make this prediction?
• Retrieving images that maximally activate a neuron [Girshick et
al. 2014]
• Finding the most influential part from the image [Zhou et al. 2016]
But, they assumed a
fixed model
Introduction
• Existing Methods
• Treat model as fixed
• Explain prediction w.r.t parameters or test input
• Our Method
• Treat model as a function of training data
• Explain prediction w.r.t the training data “most
responsible” for prediction
• How would the prediction change if we up-weighted/
modified a training point?
Influence Function
• Introduction
• Efficiently calculation
• Validation and extension
Influence Function
• the origin loss function
• optimized theta
• If we up-weighted a point z by e, new loss function
• optimized new theta
Influence Function
• We are interested in the parameter & test loss change.
• Theta change:
• Loss change:
Influence Function
• We define two influence function
• - ≈
• F(ε) = argmin =
• F(ε) ≈ F(0) + ε *
Deriving
Deriving
Deriving
• Use Taylor expansion on the right side
• F(θ)=
• F(θz) = F(θ) + (θz - θ) * F’(θ)
• F(θz) = 0
Deriving
Deriving
• Finally
Deriving other functions
Perturbing a training input
• If we change (x, y) to (x + delta, y), what will test loss
change?
• (x, y) to (x + delta, y) equals to delete (x, y) then add (x +
delta, y)
Efficiently calculation
• Two challenges:
• calculating Inverse Hessian Matrix
• calculating influence function on all training points
• n training points, p parameters
• Inverting Hessian O(np2 + p3)
• Use Conjugate gradients(refer to paper), O(np)
• Stochastic estimation(refer to paper)
Validation and Extensions
• There are some assumptions & approximation:
• model parameter minimized the loss
• the loss is twice-differentiable
• We want to check the performance of influence function
when these assumptions are violated.
Validation and Extensions
• Influence function vs leave-one-out retraining
• actually retrain a linear regression model after
removing a training point
Validation and Extensions
• Non-convexity and non-convergence
• When theta is not a minimizer, the loss change will be a
little different( refer to paper)
• Iloss non-convex
• Person’s correlation = 0.86
Validation and Extensions
• Non-differentiable losses
• Hinge loss: we can approximate using some smooth
methods
Use cases of influence functions
• Understanding model behavior
• Fixing mislabeled examples
• Adversarial training examples
• Debugging domain mismatch( refer to paper)
Understanding model behaviors
• Model1: Inception v3 with all but top layer frozen
• Model2: SVM with rbf kernel
• Task: binary image classification of fish and dog
Understanding model behaviors
• Model1: Inception v3 with all but top layer frozen
• Model2: SVM with rbf kernel
• Task: binary image classification of fish and dog
Fixing mislabeled examples
• Only have training set.
• What do we usually do?
• Find example with the largest loss
Fixing mislabeled examples
• Experiment:
• spam email data, random change 10% label
Adversarial training examples
• There exists some paper generating some adversarial test
images that are visually indistinguishable but can fool a
classifier.
• We demonstrate we can craft adversarial training images
that can flip a model’s prediction
• Basically the idea is iterating on training images on the
direction of influence function.
Adversarial training examples
• Data same as fish vs dog
• origin correctly classified 591/600 test images.
• for each test image, find only one training image and
do 100 iterations.
• 335(57%) of the testing images were flipped
• Also, attack on one training image can influence
multiple test images.
Adversarial training examples
• Data same as fish vs dog
• origin correctly classified 591/600 test images.
• for each test image, find only one training image and
do 100 iterations.
• 335(57%) of the testing images were flipped
• Also, attack on one training image can influence
multiple test images.
Thank you
code: http://bit.ly/gt-influence

More Related Content

What's hot

Decima Engine: Visibility in Horizon Zero Dawn
Decima Engine: Visibility in Horizon Zero DawnDecima Engine: Visibility in Horizon Zero Dawn
Decima Engine: Visibility in Horizon Zero DawnGuerrilla
 
Image-to-Image Translation
Image-to-Image TranslationImage-to-Image Translation
Image-to-Image TranslationJunho Kim
 
Semantic Segmentation Review
Semantic Segmentation ReviewSemantic Segmentation Review
Semantic Segmentation ReviewTakeshi Otsuka
 
Transfer Learning -- The Next Frontier for Machine Learning
Transfer Learning -- The Next Frontier for Machine LearningTransfer Learning -- The Next Frontier for Machine Learning
Transfer Learning -- The Next Frontier for Machine LearningSebastian Ruder
 
Adversarial search
Adversarial searchAdversarial search
Adversarial searchDheerendra k
 
【DL輪読会】DiffRF: Rendering-guided 3D Radiance Field Diffusion [N. Muller+ CVPR2...
【DL輪読会】DiffRF: Rendering-guided 3D Radiance Field Diffusion [N. Muller+ CVPR2...【DL輪読会】DiffRF: Rendering-guided 3D Radiance Field Diffusion [N. Muller+ CVPR2...
【DL輪読会】DiffRF: Rendering-guided 3D Radiance Field Diffusion [N. Muller+ CVPR2...Deep Learning JP
 
All about agents jade
All about agents jadeAll about agents jade
All about agents jadeAryan Rathore
 
Understanding Black-box Predictions via Influence Functions
Understanding Black-box Predictions via Influence FunctionsUnderstanding Black-box Predictions via Influence Functions
Understanding Black-box Predictions via Influence FunctionsZabir Al Nazi Nabil
 
Rendering Technologies from Crysis 3 (GDC 2013)
Rendering Technologies from Crysis 3 (GDC 2013)Rendering Technologies from Crysis 3 (GDC 2013)
Rendering Technologies from Crysis 3 (GDC 2013)Tiago Sousa
 
Transfer Learning and Domain Adaptation - Ramon Morros - UPC Barcelona 2018
Transfer Learning and Domain Adaptation - Ramon Morros - UPC Barcelona 2018Transfer Learning and Domain Adaptation - Ramon Morros - UPC Barcelona 2018
Transfer Learning and Domain Adaptation - Ramon Morros - UPC Barcelona 2018Universitat Politècnica de Catalunya
 
Artificial Intelligence 1 Planning In The Real World
Artificial Intelligence 1 Planning In The Real WorldArtificial Intelligence 1 Planning In The Real World
Artificial Intelligence 1 Planning In The Real Worldahmad bassiouny
 
Ant colony optimization
Ant colony optimizationAnt colony optimization
Ant colony optimizationJoy Dutta
 
Deep learning on mobile
Deep learning on mobileDeep learning on mobile
Deep learning on mobileAnirudh Koul
 
【DL輪読会】Novel View Synthesis with Diffusion Models
【DL輪読会】Novel View Synthesis with Diffusion Models【DL輪読会】Novel View Synthesis with Diffusion Models
【DL輪読会】Novel View Synthesis with Diffusion ModelsDeep Learning JP
 
Tutorial on Object Detection (Faster R-CNN)
Tutorial on Object Detection (Faster R-CNN)Tutorial on Object Detection (Faster R-CNN)
Tutorial on Object Detection (Faster R-CNN)Hwa Pyung Kim
 
Advancements in-tiled-rendering
Advancements in-tiled-renderingAdvancements in-tiled-rendering
Advancements in-tiled-renderingmistercteam
 
Problem Formulation in Artificial Inteligence Projects
Problem Formulation in Artificial Inteligence ProjectsProblem Formulation in Artificial Inteligence Projects
Problem Formulation in Artificial Inteligence ProjectsDr. C.V. Suresh Babu
 

What's hot (20)

Problem space
Problem spaceProblem space
Problem space
 
Decima Engine: Visibility in Horizon Zero Dawn
Decima Engine: Visibility in Horizon Zero DawnDecima Engine: Visibility in Horizon Zero Dawn
Decima Engine: Visibility in Horizon Zero Dawn
 
Image-to-Image Translation
Image-to-Image TranslationImage-to-Image Translation
Image-to-Image Translation
 
Semantic Segmentation Review
Semantic Segmentation ReviewSemantic Segmentation Review
Semantic Segmentation Review
 
Transfer Learning -- The Next Frontier for Machine Learning
Transfer Learning -- The Next Frontier for Machine LearningTransfer Learning -- The Next Frontier for Machine Learning
Transfer Learning -- The Next Frontier for Machine Learning
 
Adversarial search
Adversarial searchAdversarial search
Adversarial search
 
【DL輪読会】DiffRF: Rendering-guided 3D Radiance Field Diffusion [N. Muller+ CVPR2...
【DL輪読会】DiffRF: Rendering-guided 3D Radiance Field Diffusion [N. Muller+ CVPR2...【DL輪読会】DiffRF: Rendering-guided 3D Radiance Field Diffusion [N. Muller+ CVPR2...
【DL輪読会】DiffRF: Rendering-guided 3D Radiance Field Diffusion [N. Muller+ CVPR2...
 
All about agents jade
All about agents jadeAll about agents jade
All about agents jade
 
Understanding Black-box Predictions via Influence Functions
Understanding Black-box Predictions via Influence FunctionsUnderstanding Black-box Predictions via Influence Functions
Understanding Black-box Predictions via Influence Functions
 
Rendering Technologies from Crysis 3 (GDC 2013)
Rendering Technologies from Crysis 3 (GDC 2013)Rendering Technologies from Crysis 3 (GDC 2013)
Rendering Technologies from Crysis 3 (GDC 2013)
 
Transfer Learning and Domain Adaptation - Ramon Morros - UPC Barcelona 2018
Transfer Learning and Domain Adaptation - Ramon Morros - UPC Barcelona 2018Transfer Learning and Domain Adaptation - Ramon Morros - UPC Barcelona 2018
Transfer Learning and Domain Adaptation - Ramon Morros - UPC Barcelona 2018
 
Artificial Intelligence 1 Planning In The Real World
Artificial Intelligence 1 Planning In The Real WorldArtificial Intelligence 1 Planning In The Real World
Artificial Intelligence 1 Planning In The Real World
 
Ant colony optimization
Ant colony optimizationAnt colony optimization
Ant colony optimization
 
Deep learning on mobile
Deep learning on mobileDeep learning on mobile
Deep learning on mobile
 
画像認識4
画像認識4画像認識4
画像認識4
 
【DL輪読会】Novel View Synthesis with Diffusion Models
【DL輪読会】Novel View Synthesis with Diffusion Models【DL輪読会】Novel View Synthesis with Diffusion Models
【DL輪読会】Novel View Synthesis with Diffusion Models
 
Tutorial on Object Detection (Faster R-CNN)
Tutorial on Object Detection (Faster R-CNN)Tutorial on Object Detection (Faster R-CNN)
Tutorial on Object Detection (Faster R-CNN)
 
Transfer Learning (D2L4 Insight@DCU Machine Learning Workshop 2017)
Transfer Learning (D2L4 Insight@DCU Machine Learning Workshop 2017)Transfer Learning (D2L4 Insight@DCU Machine Learning Workshop 2017)
Transfer Learning (D2L4 Insight@DCU Machine Learning Workshop 2017)
 
Advancements in-tiled-rendering
Advancements in-tiled-renderingAdvancements in-tiled-rendering
Advancements in-tiled-rendering
 
Problem Formulation in Artificial Inteligence Projects
Problem Formulation in Artificial Inteligence ProjectsProblem Formulation in Artificial Inteligence Projects
Problem Formulation in Artificial Inteligence Projects
 

Similar to ICML2017 best paper (Understanding black box predictions via influence functions)

Linear Regression.pptx
Linear Regression.pptxLinear Regression.pptx
Linear Regression.pptxnathansel1
 
Bisection & Regual falsi methods
Bisection & Regual falsi methodsBisection & Regual falsi methods
Bisection & Regual falsi methodsDivya Bhatia
 
Week 2 - ML models and Linear Regression.pptx
Week 2 - ML models and Linear Regression.pptxWeek 2 - ML models and Linear Regression.pptx
Week 2 - ML models and Linear Regression.pptxHafizAliHummad
 
Optimization (DLAI D4L1 2017 UPC Deep Learning for Artificial Intelligence)
Optimization (DLAI D4L1 2017 UPC Deep Learning for Artificial Intelligence)Optimization (DLAI D4L1 2017 UPC Deep Learning for Artificial Intelligence)
Optimization (DLAI D4L1 2017 UPC Deep Learning for Artificial Intelligence)Universitat Politècnica de Catalunya
 
Coursera 1week
Coursera  1weekCoursera  1week
Coursera 1weekcsl9496
 
Optimization for Neural Network Training - Veronica Vilaplana - UPC Barcelona...
Optimization for Neural Network Training - Veronica Vilaplana - UPC Barcelona...Optimization for Neural Network Training - Veronica Vilaplana - UPC Barcelona...
Optimization for Neural Network Training - Veronica Vilaplana - UPC Barcelona...Universitat Politècnica de Catalunya
 
cos323_s06_lecture03_optimization.ppt
cos323_s06_lecture03_optimization.pptcos323_s06_lecture03_optimization.ppt
cos323_s06_lecture03_optimization.pptdevesh604174
 
[FLOLAC'14][scm] Functional Programming Using Haskell
[FLOLAC'14][scm] Functional Programming Using Haskell[FLOLAC'14][scm] Functional Programming Using Haskell
[FLOLAC'14][scm] Functional Programming Using HaskellFunctional Thursday
 
L1 intro2 supervised_learning
L1 intro2 supervised_learningL1 intro2 supervised_learning
L1 intro2 supervised_learningYogendra Singh
 
AP Advantage: AP Calculus
AP Advantage: AP CalculusAP Advantage: AP Calculus
AP Advantage: AP CalculusShashank Patil
 
Bartosz Milewski, “Re-discovering Monads in C++”
Bartosz Milewski, “Re-discovering Monads in C++”Bartosz Milewski, “Re-discovering Monads in C++”
Bartosz Milewski, “Re-discovering Monads in C++”Platonov Sergey
 
Reinforcement Learning and Artificial Neural Nets
Reinforcement Learning and Artificial Neural NetsReinforcement Learning and Artificial Neural Nets
Reinforcement Learning and Artificial Neural NetsPierre de Lacaze
 
Lesson3.2 a basicdifferentiationrules
Lesson3.2 a basicdifferentiationrulesLesson3.2 a basicdifferentiationrules
Lesson3.2 a basicdifferentiationrulesNico Reyes
 
Model Selection and Validation
Model Selection and ValidationModel Selection and Validation
Model Selection and Validationgmorishita
 
13Kernel_Machines.pptx
13Kernel_Machines.pptx13Kernel_Machines.pptx
13Kernel_Machines.pptxKarasuLee
 
DeepLearningLecture.pptx
DeepLearningLecture.pptxDeepLearningLecture.pptx
DeepLearningLecture.pptxssuserf07225
 
机器学习Adaboost
机器学习Adaboost机器学习Adaboost
机器学习AdaboostShocky1
 

Similar to ICML2017 best paper (Understanding black box predictions via influence functions) (20)

Linear Regression.pptx
Linear Regression.pptxLinear Regression.pptx
Linear Regression.pptx
 
Bisection & Regual falsi methods
Bisection & Regual falsi methodsBisection & Regual falsi methods
Bisection & Regual falsi methods
 
Mit6 094 iap10_lec03
Mit6 094 iap10_lec03Mit6 094 iap10_lec03
Mit6 094 iap10_lec03
 
Week 2 - ML models and Linear Regression.pptx
Week 2 - ML models and Linear Regression.pptxWeek 2 - ML models and Linear Regression.pptx
Week 2 - ML models and Linear Regression.pptx
 
Optimization (DLAI D4L1 2017 UPC Deep Learning for Artificial Intelligence)
Optimization (DLAI D4L1 2017 UPC Deep Learning for Artificial Intelligence)Optimization (DLAI D4L1 2017 UPC Deep Learning for Artificial Intelligence)
Optimization (DLAI D4L1 2017 UPC Deep Learning for Artificial Intelligence)
 
Coursera 1week
Coursera  1weekCoursera  1week
Coursera 1week
 
Optimization for Neural Network Training - Veronica Vilaplana - UPC Barcelona...
Optimization for Neural Network Training - Veronica Vilaplana - UPC Barcelona...Optimization for Neural Network Training - Veronica Vilaplana - UPC Barcelona...
Optimization for Neural Network Training - Veronica Vilaplana - UPC Barcelona...
 
cos323_s06_lecture03_optimization.ppt
cos323_s06_lecture03_optimization.pptcos323_s06_lecture03_optimization.ppt
cos323_s06_lecture03_optimization.ppt
 
[FLOLAC'14][scm] Functional Programming Using Haskell
[FLOLAC'14][scm] Functional Programming Using Haskell[FLOLAC'14][scm] Functional Programming Using Haskell
[FLOLAC'14][scm] Functional Programming Using Haskell
 
L1 intro2 supervised_learning
L1 intro2 supervised_learningL1 intro2 supervised_learning
L1 intro2 supervised_learning
 
AP Advantage: AP Calculus
AP Advantage: AP CalculusAP Advantage: AP Calculus
AP Advantage: AP Calculus
 
Bartosz Milewski, “Re-discovering Monads in C++”
Bartosz Milewski, “Re-discovering Monads in C++”Bartosz Milewski, “Re-discovering Monads in C++”
Bartosz Milewski, “Re-discovering Monads in C++”
 
Reinforcement Learning and Artificial Neural Nets
Reinforcement Learning and Artificial Neural NetsReinforcement Learning and Artificial Neural Nets
Reinforcement Learning and Artificial Neural Nets
 
Lesson3.2 a basicdifferentiationrules
Lesson3.2 a basicdifferentiationrulesLesson3.2 a basicdifferentiationrules
Lesson3.2 a basicdifferentiationrules
 
Lec3
Lec3Lec3
Lec3
 
Model Selection and Validation
Model Selection and ValidationModel Selection and Validation
Model Selection and Validation
 
13Kernel_Machines.pptx
13Kernel_Machines.pptx13Kernel_Machines.pptx
13Kernel_Machines.pptx
 
DeepLearningLecture.pptx
DeepLearningLecture.pptxDeepLearningLecture.pptx
DeepLearningLecture.pptx
 
机器学习Adaboost
机器学习Adaboost机器学习Adaboost
机器学习Adaboost
 
Note introductions of functions
Note introductions of functionsNote introductions of functions
Note introductions of functions
 

Recently uploaded

HARMONY IN THE HUMAN BEING - Unit-II UHV-2
HARMONY IN THE HUMAN BEING - Unit-II UHV-2HARMONY IN THE HUMAN BEING - Unit-II UHV-2
HARMONY IN THE HUMAN BEING - Unit-II UHV-2RajaP95
 
Architect Hassan Khalil Portfolio for 2024
Architect Hassan Khalil Portfolio for 2024Architect Hassan Khalil Portfolio for 2024
Architect Hassan Khalil Portfolio for 2024hassan khalil
 
Introduction to Microprocesso programming and interfacing.pptx
Introduction to Microprocesso programming and interfacing.pptxIntroduction to Microprocesso programming and interfacing.pptx
Introduction to Microprocesso programming and interfacing.pptxvipinkmenon1
 
Past, Present and Future of Generative AI
Past, Present and Future of Generative AIPast, Present and Future of Generative AI
Past, Present and Future of Generative AIabhishek36461
 
Sachpazis Costas: Geotechnical Engineering: A student's Perspective Introduction
Sachpazis Costas: Geotechnical Engineering: A student's Perspective IntroductionSachpazis Costas: Geotechnical Engineering: A student's Perspective Introduction
Sachpazis Costas: Geotechnical Engineering: A student's Perspective IntroductionDr.Costas Sachpazis
 
VICTOR MAESTRE RAMIREZ - Planetary Defender on NASA's Double Asteroid Redirec...
VICTOR MAESTRE RAMIREZ - Planetary Defender on NASA's Double Asteroid Redirec...VICTOR MAESTRE RAMIREZ - Planetary Defender on NASA's Double Asteroid Redirec...
VICTOR MAESTRE RAMIREZ - Planetary Defender on NASA's Double Asteroid Redirec...VICTOR MAESTRE RAMIREZ
 
OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...
OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...
OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...Soham Mondal
 
Internship report on mechanical engineering
Internship report on mechanical engineeringInternship report on mechanical engineering
Internship report on mechanical engineeringmalavadedarshan25
 
main PPT.pptx of girls hostel security using rfid
main PPT.pptx of girls hostel security using rfidmain PPT.pptx of girls hostel security using rfid
main PPT.pptx of girls hostel security using rfidNikhilNagaraju
 
APPLICATIONS-AC/DC DRIVES-OPERATING CHARACTERISTICS
APPLICATIONS-AC/DC DRIVES-OPERATING CHARACTERISTICSAPPLICATIONS-AC/DC DRIVES-OPERATING CHARACTERISTICS
APPLICATIONS-AC/DC DRIVES-OPERATING CHARACTERISTICSKurinjimalarL3
 
Decoding Kotlin - Your guide to solving the mysterious in Kotlin.pptx
Decoding Kotlin - Your guide to solving the mysterious in Kotlin.pptxDecoding Kotlin - Your guide to solving the mysterious in Kotlin.pptx
Decoding Kotlin - Your guide to solving the mysterious in Kotlin.pptxJoão Esperancinha
 
Biology for Computer Engineers Course Handout.pptx
Biology for Computer Engineers Course Handout.pptxBiology for Computer Engineers Course Handout.pptx
Biology for Computer Engineers Course Handout.pptxDeepakSakkari2
 
Call Girls Narol 7397865700 Independent Call Girls
Call Girls Narol 7397865700 Independent Call GirlsCall Girls Narol 7397865700 Independent Call Girls
Call Girls Narol 7397865700 Independent Call Girlsssuser7cb4ff
 
Gfe Mayur Vihar Call Girls Service WhatsApp -> 9999965857 Available 24x7 ^ De...
Gfe Mayur Vihar Call Girls Service WhatsApp -> 9999965857 Available 24x7 ^ De...Gfe Mayur Vihar Call Girls Service WhatsApp -> 9999965857 Available 24x7 ^ De...
Gfe Mayur Vihar Call Girls Service WhatsApp -> 9999965857 Available 24x7 ^ De...srsj9000
 
Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...
Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...
Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...Dr.Costas Sachpazis
 
(ANVI) Koregaon Park Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
(ANVI) Koregaon Park Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...(ANVI) Koregaon Park Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
(ANVI) Koregaon Park Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...ranjana rawat
 
VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130
VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130
VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130Suhani Kapoor
 
microprocessor 8085 and its interfacing
microprocessor 8085  and its interfacingmicroprocessor 8085  and its interfacing
microprocessor 8085 and its interfacingjaychoudhary37
 

Recently uploaded (20)

HARMONY IN THE HUMAN BEING - Unit-II UHV-2
HARMONY IN THE HUMAN BEING - Unit-II UHV-2HARMONY IN THE HUMAN BEING - Unit-II UHV-2
HARMONY IN THE HUMAN BEING - Unit-II UHV-2
 
Architect Hassan Khalil Portfolio for 2024
Architect Hassan Khalil Portfolio for 2024Architect Hassan Khalil Portfolio for 2024
Architect Hassan Khalil Portfolio for 2024
 
Introduction to Microprocesso programming and interfacing.pptx
Introduction to Microprocesso programming and interfacing.pptxIntroduction to Microprocesso programming and interfacing.pptx
Introduction to Microprocesso programming and interfacing.pptx
 
9953056974 Call Girls In South Ex, Escorts (Delhi) NCR.pdf
9953056974 Call Girls In South Ex, Escorts (Delhi) NCR.pdf9953056974 Call Girls In South Ex, Escorts (Delhi) NCR.pdf
9953056974 Call Girls In South Ex, Escorts (Delhi) NCR.pdf
 
Past, Present and Future of Generative AI
Past, Present and Future of Generative AIPast, Present and Future of Generative AI
Past, Present and Future of Generative AI
 
Sachpazis Costas: Geotechnical Engineering: A student's Perspective Introduction
Sachpazis Costas: Geotechnical Engineering: A student's Perspective IntroductionSachpazis Costas: Geotechnical Engineering: A student's Perspective Introduction
Sachpazis Costas: Geotechnical Engineering: A student's Perspective Introduction
 
VICTOR MAESTRE RAMIREZ - Planetary Defender on NASA's Double Asteroid Redirec...
VICTOR MAESTRE RAMIREZ - Planetary Defender on NASA's Double Asteroid Redirec...VICTOR MAESTRE RAMIREZ - Planetary Defender on NASA's Double Asteroid Redirec...
VICTOR MAESTRE RAMIREZ - Planetary Defender on NASA's Double Asteroid Redirec...
 
OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...
OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...
OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...
 
Internship report on mechanical engineering
Internship report on mechanical engineeringInternship report on mechanical engineering
Internship report on mechanical engineering
 
main PPT.pptx of girls hostel security using rfid
main PPT.pptx of girls hostel security using rfidmain PPT.pptx of girls hostel security using rfid
main PPT.pptx of girls hostel security using rfid
 
APPLICATIONS-AC/DC DRIVES-OPERATING CHARACTERISTICS
APPLICATIONS-AC/DC DRIVES-OPERATING CHARACTERISTICSAPPLICATIONS-AC/DC DRIVES-OPERATING CHARACTERISTICS
APPLICATIONS-AC/DC DRIVES-OPERATING CHARACTERISTICS
 
Decoding Kotlin - Your guide to solving the mysterious in Kotlin.pptx
Decoding Kotlin - Your guide to solving the mysterious in Kotlin.pptxDecoding Kotlin - Your guide to solving the mysterious in Kotlin.pptx
Decoding Kotlin - Your guide to solving the mysterious in Kotlin.pptx
 
Biology for Computer Engineers Course Handout.pptx
Biology for Computer Engineers Course Handout.pptxBiology for Computer Engineers Course Handout.pptx
Biology for Computer Engineers Course Handout.pptx
 
Call Girls Narol 7397865700 Independent Call Girls
Call Girls Narol 7397865700 Independent Call GirlsCall Girls Narol 7397865700 Independent Call Girls
Call Girls Narol 7397865700 Independent Call Girls
 
Gfe Mayur Vihar Call Girls Service WhatsApp -> 9999965857 Available 24x7 ^ De...
Gfe Mayur Vihar Call Girls Service WhatsApp -> 9999965857 Available 24x7 ^ De...Gfe Mayur Vihar Call Girls Service WhatsApp -> 9999965857 Available 24x7 ^ De...
Gfe Mayur Vihar Call Girls Service WhatsApp -> 9999965857 Available 24x7 ^ De...
 
Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...
Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...
Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...
 
young call girls in Rajiv Chowk🔝 9953056974 🔝 Delhi escort Service
young call girls in Rajiv Chowk🔝 9953056974 🔝 Delhi escort Serviceyoung call girls in Rajiv Chowk🔝 9953056974 🔝 Delhi escort Service
young call girls in Rajiv Chowk🔝 9953056974 🔝 Delhi escort Service
 
(ANVI) Koregaon Park Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
(ANVI) Koregaon Park Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...(ANVI) Koregaon Park Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
(ANVI) Koregaon Park Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
 
VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130
VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130
VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130
 
microprocessor 8085 and its interfacing
microprocessor 8085  and its interfacingmicroprocessor 8085  and its interfacing
microprocessor 8085 and its interfacing
 

ICML2017 best paper (Understanding black box predictions via influence functions)

  • 1. Understanding black-box predictions via influence functions XIE Ruiming
  • 2. Outline • Background • Taylor's Formula • Newton's Method • Introduction • Influence Function • Definition • Efficiently Calculating Influence • Validation and Extensions • Use cases of influence function
  • 3. Background : Taylor • Taylor's theorem gives an approximation of a k-times differentiable function around a given point by a k-th order Taylor polynomial • Linear approximation • Quadratic approximation
  • 4. Background : Newton • Find x:F(x) = 0 through iteration. • Recall Taylor’s Formula • F(a) ≈ F(Xn) + F’(Xn)(a – Xn) • Set F(a) = 0, get a = Xn – F(Xn)/F’(Xn) • Newton in optimizing • X = argmin(F(x)), then F’(X) = 0 • Doing Newton’s method with F’(X) • Xn+1 = Xn – F’(Xn)/F’’(Xn)
  • 5. Background : Newton • X = parameters
  • 7. Introduction • Why did the model make this prediction?
  • 8. Introduction • Why did the model make this prediction? • Retrieving images that maximally activate a neuron [Girshick et al. 2014]
  • 9. Introduction • Why did the model make this prediction? • Retrieving images that maximally activate a neuron [Girshick et al. 2014] • Finding the most influential part from the image [Zhou et al. 2016]
  • 10. Introduction • Why did the model make this prediction? • Retrieving images that maximally activate a neuron [Girshick et al. 2014] • Finding the most influential part from the image [Zhou et al. 2016] But, they assumed a fixed model
  • 11. Introduction • Existing Methods • Treat model as fixed • Explain prediction w.r.t parameters or test input • Our Method • Treat model as a function of training data • Explain prediction w.r.t the training data “most responsible” for prediction • How would the prediction change if we up-weighted/ modified a training point?
  • 12. Influence Function • Introduction • Efficiently calculation • Validation and extension
  • 13. Influence Function • the origin loss function • optimized theta • If we up-weighted a point z by e, new loss function • optimized new theta
  • 14. Influence Function • We are interested in the parameter & test loss change. • Theta change: • Loss change:
  • 15. Influence Function • We define two influence function • - ≈ • F(ε) = argmin = • F(ε) ≈ F(0) + ε *
  • 18. Deriving • Use Taylor expansion on the right side • F(θ)= • F(θz) = F(θ) + (θz - θ) * F’(θ) • F(θz) = 0
  • 22. Perturbing a training input • If we change (x, y) to (x + delta, y), what will test loss change? • (x, y) to (x + delta, y) equals to delete (x, y) then add (x + delta, y)
  • 23. Efficiently calculation • Two challenges: • calculating Inverse Hessian Matrix • calculating influence function on all training points • n training points, p parameters • Inverting Hessian O(np2 + p3) • Use Conjugate gradients(refer to paper), O(np) • Stochastic estimation(refer to paper)
  • 24. Validation and Extensions • There are some assumptions & approximation: • model parameter minimized the loss • the loss is twice-differentiable • We want to check the performance of influence function when these assumptions are violated.
  • 25. Validation and Extensions • Influence function vs leave-one-out retraining • actually retrain a linear regression model after removing a training point
  • 26. Validation and Extensions • Non-convexity and non-convergence • When theta is not a minimizer, the loss change will be a little different( refer to paper) • Iloss non-convex • Person’s correlation = 0.86
  • 27. Validation and Extensions • Non-differentiable losses • Hinge loss: we can approximate using some smooth methods
  • 28. Use cases of influence functions • Understanding model behavior • Fixing mislabeled examples • Adversarial training examples • Debugging domain mismatch( refer to paper)
  • 29. Understanding model behaviors • Model1: Inception v3 with all but top layer frozen • Model2: SVM with rbf kernel • Task: binary image classification of fish and dog
  • 30. Understanding model behaviors • Model1: Inception v3 with all but top layer frozen • Model2: SVM with rbf kernel • Task: binary image classification of fish and dog
  • 31. Fixing mislabeled examples • Only have training set. • What do we usually do? • Find example with the largest loss
  • 32. Fixing mislabeled examples • Experiment: • spam email data, random change 10% label
  • 33. Adversarial training examples • There exists some paper generating some adversarial test images that are visually indistinguishable but can fool a classifier. • We demonstrate we can craft adversarial training images that can flip a model’s prediction • Basically the idea is iterating on training images on the direction of influence function.
  • 34. Adversarial training examples • Data same as fish vs dog • origin correctly classified 591/600 test images. • for each test image, find only one training image and do 100 iterations. • 335(57%) of the testing images were flipped • Also, attack on one training image can influence multiple test images.
  • 35. Adversarial training examples • Data same as fish vs dog • origin correctly classified 591/600 test images. • for each test image, find only one training image and do 100 iterations. • 335(57%) of the testing images were flipped • Also, attack on one training image can influence multiple test images.