SlideShare uma empresa Scribd logo
1 de 47
Baixar para ler offline
2
The good and the evil
3
The good and the evil
4
Mélanie DUCOFFE
PhD., Research Data
Scientist
@mducoffe
melanie.ducoffe@airbus.com
Deep Learning, AI explainability
About me !
5
Clément DUFFAU
PhD., Lead DevOps
@clement0210
clement.duffau@stack-labs.com
Automation, verification & validation,
assurance quality, safety
About me !
6
Summary
Artificial Intelligence and ethics
Data selection biases
In production biases
Train interpretation biases
7
Job posts gathered into a single position: prone to many biases. Data biases and Model biases
What do you think of combining : architect + dev+ testing + maintenance ?
The nature of AI into the work market
Data Scientist
Data Analyst
Data Engineer
Machine Learning Scientist
Machine Learning Engineer
Statistician
Find, clean and organize data for companies
Transform and manipulate large datasets to suit the analysis of companies
Perform batch processing or real time processing on gathered/stored data
Research new data approaches and algorithms
Create data funnels and deliver software solutions
Interpret, analyze and report statistical information
8
● Bias (in statistics) : “the difference between the expectation of the sample estimator and the true population value, which
reduces the representativeness of the estimator by systematically distorting it”
● “The big takeaway is that we don’t know what we don’t know,” (Alice Popejoy)
What do you mean about biases ?
9
Can we
anticipate
biases ?
10
Ethics in AI
11
● MIT study decision-making of self-driving car on
killing scenario
● Scenario with
○ high/low level of education
○ young/old
○ male/female
○ pets
○ traffic signal respect
http://moralmachine.mit.edu/
The Moral Machine
12
● Results “rank” you on
○ number of death
○ law respect
○ gender,age,health, ...
● Need ethical choices at the government,
insurance ? manufacturer ? passengers ?
The Moral Machine
13
● 3 different clusters
● Cultural differences
○ USA/Europe kill the oldest vs Japon saves the oldest
○ Colombia saves high educated people vs Finland doesn’t matter
○ South america/France saves women
The Moral Machine experiment, Edmond Awad et al. (2018), Nature
The Moral Machine
14
Ethics in clinical investigations
● Technical committee to review scientific foundations and safety (in France, ANSM)
● Ethics committee on animal AND humans (in France, CPP)
○ Composed by medical professional and citizen
○ Review application form on
■ benefit/risk
■ information quality
■ resources to conduct the study
■ patient recruitment process
■ Patient consent modality
○ Follow the experiment process
15
Applying ethics committee in AI at Google
● Advanced Technology External Advisory Council (ATEAC)
○ Ensure the white paper principles for AI at Google
■ Be socially beneficial
■ Avoid creating or reinforcing unfair bias
■ Be built and tested for safety
■ Be accountable to people
■ Incorporate privacy design principles
■ Uphold high standards of scientific excellence
■ Be made available for uses that accord with these principles
● Dissolved 1 week after creation
○ Ethics of some attendees were discussed …
○ Questioning themselves of the needs to represent every part of the society
16
Data
biases Data
biases
17
● In the 90s, “I was breaking all of [my classmates’] facial-recognition software
because apparently all the pictures they were taking were of people with
significantly less melanin than I have” (Charles Isbell)
● In 2015, the “gorilla mistake” in Google Photos
● 25 years, not the same learning model at all, but the same root cause
Face recognition from the 90s to nowadays
18
● In 2014, Amazon develops an IA to “find” key success factor and hire these people
● Train on their 10 last years hired people
● 89% of the engineering workforce is male
● At the end, you have sexist recruitment AI !
● Representative data inside the company by not inside the society …
AI recruitment by Amazon
19
Biases in the Features
Outcome outputFeatures
Model
Diploma Gender Hobbies
Nationality University ...
HIRED ?
Model
Outcome output
Features
● Lack of an appropriate set of features
● Lack of an appropriate dataset
● Imbalanced dataset or bias in the output
● Unawareness: remove sensitive features from your
data
20
Tutorial: your first bias detector
Survival of passengers on the Titanic
● Decision Tree
● Leaves = class labels
● Nodes = splitting/conjunctions of features
● Important nodes = less deep, lot of observations
● Many Framework coexist:
FairML, “What If”, IBM Bias Assessment
● ‘No universal solution’: combine them
Summarize: Your chance of survival were
good iff:
- you were a woman
- you were a young boy with few siblings
21
● Bad advice pointed out by IBM in internal in 2017
○ ex : Suggest a cancer patient with severe bleeding be given a drug that could cause the bleeding to worsen
● Started off by using real patient data
● Fed it with hypothetical data
● “Synthetic cases allow you to treat and train Watson on a variety of patient variables and conditions that might not be
present in random patient samples, but are important to treatment recommendations” (Edward Barbani)
● Pointed out the difficulties to collect representative data in medical
Unsafe medical recommendations by IBM Watson
22
Data Augmentation
● Historically for images: rotation, flipping, adding noise…
● Object detection models : performance loss on corrupted images
● CNNs generalize poorly to novel distortion types, despite being trained on a variety of other distortions
23
Generating new data with GANS
source: Medium, “Generative Adversarial with Maths” by Madhu
Sanjeevi
Discriminator
Generator
Generator
24
Understanding biases and Generalization in Deep
Generative Models
Can we learn the distribution of data with GANS ?
What are the inductive biases of Deep Generative models ?
● Unbiased and consistent density estimation impossible
● Inductive biases
● Similar cognitive bias as humans: numerosity
● Weber’s law: relative change (ratio)
True data Generated data
25
GAN-based Data Augmentation Perpetuates biases
26
27
Training
biases Data
biases
Model
biases
28
Group Fairness
● Demographic Parity
○ ex : gender independent
● Equal odds
○ ex : take into account the reality
statistics in one side
● Equal opportunity
○ ex : take into account the reality
statistics in both side
29
AI explainability
30
Linear explanation
DOG
CAT
31
Black box models: non linear decision functions
DOG
CAT
WHY
32
LIME: Zooming !
create neighbors and
deduce a linear
classifier
DOG
CAT
DOG
CAT
33
Is it working ?
Examples of Explanations are usually cherry-picked !
34
Individual Fairness for Explainability
The fact that 2 resulting saliency maps are different is fundamentally due to the network itself being fragile to
such perturbations
35
Autonomous Vehicle Visualization at Uber
36
In production
biases Data
biases
Model
biases
Prod
biases
37
● Experienced lane recognition algorithm
● Doesn’t expect that 3 stickers break the robustness
● Hacking the AI with thinks that doesn’t matter humans
Tesla autonomous driving hacked with stickers
38
Hack: Adversarial examples
39
Hacking tutorial: design your own adversarial example
CAT DOG CAT DOG
if B is predicted as
a cat, retry to cross
the decision
boundary with B !
Linear Model NON Linear Model
A
A
B
B
40
Hacking Tutorial: I don’t have access to the model
This is not a
car !
This is not a
car !
41
Defending against adversarial attacks ?!
SUCCESS
FAILURES
2013
Discovery of
adversarial examples
2015
Fast Gradient
Sign
2015
DeepFool
2016
Carlini Wagner
2019
Unforeseen
Attack
2013
Adversarial Training
BRUTE FORCE: training with
adv
2015
Defensive Distillation
output probabilities rather than hard
decisions
2016
JPEG compression
against adv
42
Why is it hard to defend against adversarial examples?
● Adaptative: block one type of attacks but leaves vulnerability open to an attacker who knows the defense being used
● Hard to defend because hard to understand the theory ?
43
Adversarial Examples are not Bugs, they are
features
BIRD
DOG
CAT
“Bird”
“Dog”
“Cat”
create the label-target adversarial dataset
DOG
CAT
BIRD
“Dog”
“Cat”
“Bird”
44
Semantic features ?
BIRD
CAT
“Bird”
“Cat”
45
Conclusion
Data
Biases
Model
Biases
Explanation
Prod
Biases
Adv
Examples
If no one cares, it is highly likely that the next person who suffers from biased treatment is one of us
46
What about in the future ?
● Art project exposed racial bias in the biggest image dataset
ImageNet in Sept. 2019
● ImageNet will remove 600,000 images
● Number of publications per year with the name Imagenet in
the abstract…
47
THANKS!

Mais conteúdo relacionado

Semelhante a When the AIs failures send us back to our own societal biases

Measures and mismeasures of algorithmic fairness
Measures and mismeasures of algorithmic fairnessMeasures and mismeasures of algorithmic fairness
Measures and mismeasures of algorithmic fairnessManojit Nandi
 
Ai demystified for HR and TA leaders
Ai demystified for HR and TA leadersAi demystified for HR and TA leaders
Ai demystified for HR and TA leadersAntonia Macrides
 
AI in the Real World: Challenges, and Risks and how to handle them?
AI in the Real World: Challenges, and Risks and how to handle them?AI in the Real World: Challenges, and Risks and how to handle them?
AI in the Real World: Challenges, and Risks and how to handle them?Srinath Perera
 
Responsible AI in Industry (Tutorials at AAAI 2021, FAccT 2021, and WWW 2021)
Responsible AI in Industry (Tutorials at AAAI 2021, FAccT 2021, and WWW 2021)Responsible AI in Industry (Tutorials at AAAI 2021, FAccT 2021, and WWW 2021)
Responsible AI in Industry (Tutorials at AAAI 2021, FAccT 2021, and WWW 2021)Krishnaram Kenthapadi
 
Improving How We Deliver Machine Learning Models (XCONF 2019)
Improving How We Deliver Machine Learning Models (XCONF 2019)Improving How We Deliver Machine Learning Models (XCONF 2019)
Improving How We Deliver Machine Learning Models (XCONF 2019)David Tan
 
Machine Learning
Machine LearningMachine Learning
Machine LearningVivek Garg
 
Scale your Testing and Quality with Automation Engineering and ML - Carlos Ki...
Scale your Testing and Quality with Automation Engineering and ML - Carlos Ki...Scale your Testing and Quality with Automation Engineering and ML - Carlos Ki...
Scale your Testing and Quality with Automation Engineering and ML - Carlos Ki...QA or the Highway
 
Responsible AI in Industry (ICML 2021 Tutorial)
Responsible AI in Industry (ICML 2021 Tutorial)Responsible AI in Industry (ICML 2021 Tutorial)
Responsible AI in Industry (ICML 2021 Tutorial)Krishnaram Kenthapadi
 
Ml topic1 a
Ml topic1 aMl topic1 a
Ml topic1 abosycs1
 
Human in the loop: Bayesian Rules Enabling Explainable AI
Human in the loop: Bayesian Rules Enabling Explainable AIHuman in the loop: Bayesian Rules Enabling Explainable AI
Human in the loop: Bayesian Rules Enabling Explainable AIPramit Choudhary
 
Defining the boundary for AI research in Intelligent Systems Dec 2021
Defining the boundary for AI research in Intelligent Systems Dec  2021Defining the boundary for AI research in Intelligent Systems Dec  2021
Defining the boundary for AI research in Intelligent Systems Dec 2021Parasuram Balasubramanian
 
“Responsible AI: Tools and Frameworks for Developing AI Solutions,” a Present...
“Responsible AI: Tools and Frameworks for Developing AI Solutions,” a Present...“Responsible AI: Tools and Frameworks for Developing AI Solutions,” a Present...
“Responsible AI: Tools and Frameworks for Developing AI Solutions,” a Present...Edge AI and Vision Alliance
 
Lecture1 introduction to machine learning
Lecture1 introduction to machine learningLecture1 introduction to machine learning
Lecture1 introduction to machine learningUmmeSalmaM1
 
Responsible AI in Industry: Practical Challenges and Lessons Learned
Responsible AI in Industry: Practical Challenges and Lessons LearnedResponsible AI in Industry: Practical Challenges and Lessons Learned
Responsible AI in Industry: Practical Challenges and Lessons LearnedKrishnaram Kenthapadi
 
Adversarial Attacks and Defense
Adversarial Attacks and DefenseAdversarial Attacks and Defense
Adversarial Attacks and DefenseKishor Datta Gupta
 
Machine Learning: Opening the Pandora's Box - Dhiana Deva @ QCon São Paulo 2019
Machine Learning: Opening the Pandora's Box - Dhiana Deva @ QCon São Paulo 2019Machine Learning: Opening the Pandora's Box - Dhiana Deva @ QCon São Paulo 2019
Machine Learning: Opening the Pandora's Box - Dhiana Deva @ QCon São Paulo 2019Dhiana Deva
 
Making fashion recommendations with human-in-the-loop machine learning
Making fashion recommendations with human-in-the-loop machine learningMaking fashion recommendations with human-in-the-loop machine learning
Making fashion recommendations with human-in-the-loop machine learningBrad Klingenberg
 

Semelhante a When the AIs failures send us back to our own societal biases (20)

Measures and mismeasures of algorithmic fairness
Measures and mismeasures of algorithmic fairnessMeasures and mismeasures of algorithmic fairness
Measures and mismeasures of algorithmic fairness
 
Ai demystified for HR and TA leaders
Ai demystified for HR and TA leadersAi demystified for HR and TA leaders
Ai demystified for HR and TA leaders
 
AI in the Real World: Challenges, and Risks and how to handle them?
AI in the Real World: Challenges, and Risks and how to handle them?AI in the Real World: Challenges, and Risks and how to handle them?
AI in the Real World: Challenges, and Risks and how to handle them?
 
Responsible AI in Industry (Tutorials at AAAI 2021, FAccT 2021, and WWW 2021)
Responsible AI in Industry (Tutorials at AAAI 2021, FAccT 2021, and WWW 2021)Responsible AI in Industry (Tutorials at AAAI 2021, FAccT 2021, and WWW 2021)
Responsible AI in Industry (Tutorials at AAAI 2021, FAccT 2021, and WWW 2021)
 
Improving How We Deliver Machine Learning Models (XCONF 2019)
Improving How We Deliver Machine Learning Models (XCONF 2019)Improving How We Deliver Machine Learning Models (XCONF 2019)
Improving How We Deliver Machine Learning Models (XCONF 2019)
 
Machine Learning
Machine LearningMachine Learning
Machine Learning
 
Debugging AI
Debugging AIDebugging AI
Debugging AI
 
Ml masterclass
Ml masterclassMl masterclass
Ml masterclass
 
Scale your Testing and Quality with Automation Engineering and ML - Carlos Ki...
Scale your Testing and Quality with Automation Engineering and ML - Carlos Ki...Scale your Testing and Quality with Automation Engineering and ML - Carlos Ki...
Scale your Testing and Quality with Automation Engineering and ML - Carlos Ki...
 
Responsible AI in Industry (ICML 2021 Tutorial)
Responsible AI in Industry (ICML 2021 Tutorial)Responsible AI in Industry (ICML 2021 Tutorial)
Responsible AI in Industry (ICML 2021 Tutorial)
 
Ml topic1 a
Ml topic1 aMl topic1 a
Ml topic1 a
 
Human in the loop: Bayesian Rules Enabling Explainable AI
Human in the loop: Bayesian Rules Enabling Explainable AIHuman in the loop: Bayesian Rules Enabling Explainable AI
Human in the loop: Bayesian Rules Enabling Explainable AI
 
Defining the boundary for AI research in Intelligent Systems Dec 2021
Defining the boundary for AI research in Intelligent Systems Dec  2021Defining the boundary for AI research in Intelligent Systems Dec  2021
Defining the boundary for AI research in Intelligent Systems Dec 2021
 
“Responsible AI: Tools and Frameworks for Developing AI Solutions,” a Present...
“Responsible AI: Tools and Frameworks for Developing AI Solutions,” a Present...“Responsible AI: Tools and Frameworks for Developing AI Solutions,” a Present...
“Responsible AI: Tools and Frameworks for Developing AI Solutions,” a Present...
 
Lecture1 introduction to machine learning
Lecture1 introduction to machine learningLecture1 introduction to machine learning
Lecture1 introduction to machine learning
 
Responsible AI in Industry: Practical Challenges and Lessons Learned
Responsible AI in Industry: Practical Challenges and Lessons LearnedResponsible AI in Industry: Practical Challenges and Lessons Learned
Responsible AI in Industry: Practical Challenges and Lessons Learned
 
Adversarial Attacks and Defense
Adversarial Attacks and DefenseAdversarial Attacks and Defense
Adversarial Attacks and Defense
 
Machine Learning: Opening the Pandora's Box - Dhiana Deva @ QCon São Paulo 2019
Machine Learning: Opening the Pandora's Box - Dhiana Deva @ QCon São Paulo 2019Machine Learning: Opening the Pandora's Box - Dhiana Deva @ QCon São Paulo 2019
Machine Learning: Opening the Pandora's Box - Dhiana Deva @ QCon São Paulo 2019
 
Lec 01
Lec 01Lec 01
Lec 01
 
Making fashion recommendations with human-in-the-loop machine learning
Making fashion recommendations with human-in-the-loop machine learningMaking fashion recommendations with human-in-the-loop machine learning
Making fashion recommendations with human-in-the-loop machine learning
 

Último

VIP Call Girls Palanpur 7001035870 Whatsapp Number, 24/07 Booking
VIP Call Girls Palanpur 7001035870 Whatsapp Number, 24/07 BookingVIP Call Girls Palanpur 7001035870 Whatsapp Number, 24/07 Booking
VIP Call Girls Palanpur 7001035870 Whatsapp Number, 24/07 Bookingdharasingh5698
 
Call Girls Wakad Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Wakad Call Me 7737669865 Budget Friendly No Advance BookingCall Girls Wakad Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Wakad Call Me 7737669865 Budget Friendly No Advance Bookingroncy bisnoi
 
Generative AI or GenAI technology based PPT
Generative AI or GenAI technology based PPTGenerative AI or GenAI technology based PPT
Generative AI or GenAI technology based PPTbhaskargani46
 
Unit 1 - Soil Classification and Compaction.pdf
Unit 1 - Soil Classification and Compaction.pdfUnit 1 - Soil Classification and Compaction.pdf
Unit 1 - Soil Classification and Compaction.pdfRagavanV2
 
Call Girls Walvekar Nagar Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Walvekar Nagar Call Me 7737669865 Budget Friendly No Advance BookingCall Girls Walvekar Nagar Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Walvekar Nagar Call Me 7737669865 Budget Friendly No Advance Bookingroncy bisnoi
 
KubeKraft presentation @CloudNativeHooghly
KubeKraft presentation @CloudNativeHooghlyKubeKraft presentation @CloudNativeHooghly
KubeKraft presentation @CloudNativeHooghlysanyuktamishra911
 
Top Rated Pune Call Girls Budhwar Peth ⟟ 6297143586 ⟟ Call Me For Genuine Se...
Top Rated  Pune Call Girls Budhwar Peth ⟟ 6297143586 ⟟ Call Me For Genuine Se...Top Rated  Pune Call Girls Budhwar Peth ⟟ 6297143586 ⟟ Call Me For Genuine Se...
Top Rated Pune Call Girls Budhwar Peth ⟟ 6297143586 ⟟ Call Me For Genuine Se...Call Girls in Nagpur High Profile
 
data_management_and _data_science_cheat_sheet.pdf
data_management_and _data_science_cheat_sheet.pdfdata_management_and _data_science_cheat_sheet.pdf
data_management_and _data_science_cheat_sheet.pdfJiananWang21
 
The Most Attractive Pune Call Girls Budhwar Peth 8250192130 Will You Miss Thi...
The Most Attractive Pune Call Girls Budhwar Peth 8250192130 Will You Miss Thi...The Most Attractive Pune Call Girls Budhwar Peth 8250192130 Will You Miss Thi...
The Most Attractive Pune Call Girls Budhwar Peth 8250192130 Will You Miss Thi...ranjana rawat
 
Thermal Engineering -unit - III & IV.ppt
Thermal Engineering -unit - III & IV.pptThermal Engineering -unit - III & IV.ppt
Thermal Engineering -unit - III & IV.pptDineshKumar4165
 
Booking open Available Pune Call Girls Pargaon 6297143586 Call Hot Indian Gi...
Booking open Available Pune Call Girls Pargaon  6297143586 Call Hot Indian Gi...Booking open Available Pune Call Girls Pargaon  6297143586 Call Hot Indian Gi...
Booking open Available Pune Call Girls Pargaon 6297143586 Call Hot Indian Gi...Call Girls in Nagpur High Profile
 
Online banking management system project.pdf
Online banking management system project.pdfOnline banking management system project.pdf
Online banking management system project.pdfKamal Acharya
 
Intze Overhead Water Tank Design by Working Stress - IS Method.pdf
Intze Overhead Water Tank  Design by Working Stress - IS Method.pdfIntze Overhead Water Tank  Design by Working Stress - IS Method.pdf
Intze Overhead Water Tank Design by Working Stress - IS Method.pdfSuman Jyoti
 
Double Revolving field theory-how the rotor develops torque
Double Revolving field theory-how the rotor develops torqueDouble Revolving field theory-how the rotor develops torque
Double Revolving field theory-how the rotor develops torqueBhangaleSonal
 
notes on Evolution Of Analytic Scalability.ppt
notes on Evolution Of Analytic Scalability.pptnotes on Evolution Of Analytic Scalability.ppt
notes on Evolution Of Analytic Scalability.pptMsecMca
 
ONLINE FOOD ORDER SYSTEM PROJECT REPORT.pdf
ONLINE FOOD ORDER SYSTEM PROJECT REPORT.pdfONLINE FOOD ORDER SYSTEM PROJECT REPORT.pdf
ONLINE FOOD ORDER SYSTEM PROJECT REPORT.pdfKamal Acharya
 
CCS335 _ Neural Networks and Deep Learning Laboratory_Lab Complete Record
CCS335 _ Neural Networks and Deep Learning Laboratory_Lab Complete RecordCCS335 _ Neural Networks and Deep Learning Laboratory_Lab Complete Record
CCS335 _ Neural Networks and Deep Learning Laboratory_Lab Complete RecordAsst.prof M.Gokilavani
 
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXssuser89054b
 

Último (20)

VIP Call Girls Palanpur 7001035870 Whatsapp Number, 24/07 Booking
VIP Call Girls Palanpur 7001035870 Whatsapp Number, 24/07 BookingVIP Call Girls Palanpur 7001035870 Whatsapp Number, 24/07 Booking
VIP Call Girls Palanpur 7001035870 Whatsapp Number, 24/07 Booking
 
Call Girls Wakad Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Wakad Call Me 7737669865 Budget Friendly No Advance BookingCall Girls Wakad Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Wakad Call Me 7737669865 Budget Friendly No Advance Booking
 
Water Industry Process Automation & Control Monthly - April 2024
Water Industry Process Automation & Control Monthly - April 2024Water Industry Process Automation & Control Monthly - April 2024
Water Industry Process Automation & Control Monthly - April 2024
 
Generative AI or GenAI technology based PPT
Generative AI or GenAI technology based PPTGenerative AI or GenAI technology based PPT
Generative AI or GenAI technology based PPT
 
Unit 1 - Soil Classification and Compaction.pdf
Unit 1 - Soil Classification and Compaction.pdfUnit 1 - Soil Classification and Compaction.pdf
Unit 1 - Soil Classification and Compaction.pdf
 
Call Girls Walvekar Nagar Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Walvekar Nagar Call Me 7737669865 Budget Friendly No Advance BookingCall Girls Walvekar Nagar Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Walvekar Nagar Call Me 7737669865 Budget Friendly No Advance Booking
 
KubeKraft presentation @CloudNativeHooghly
KubeKraft presentation @CloudNativeHooghlyKubeKraft presentation @CloudNativeHooghly
KubeKraft presentation @CloudNativeHooghly
 
Top Rated Pune Call Girls Budhwar Peth ⟟ 6297143586 ⟟ Call Me For Genuine Se...
Top Rated  Pune Call Girls Budhwar Peth ⟟ 6297143586 ⟟ Call Me For Genuine Se...Top Rated  Pune Call Girls Budhwar Peth ⟟ 6297143586 ⟟ Call Me For Genuine Se...
Top Rated Pune Call Girls Budhwar Peth ⟟ 6297143586 ⟟ Call Me For Genuine Se...
 
data_management_and _data_science_cheat_sheet.pdf
data_management_and _data_science_cheat_sheet.pdfdata_management_and _data_science_cheat_sheet.pdf
data_management_and _data_science_cheat_sheet.pdf
 
The Most Attractive Pune Call Girls Budhwar Peth 8250192130 Will You Miss Thi...
The Most Attractive Pune Call Girls Budhwar Peth 8250192130 Will You Miss Thi...The Most Attractive Pune Call Girls Budhwar Peth 8250192130 Will You Miss Thi...
The Most Attractive Pune Call Girls Budhwar Peth 8250192130 Will You Miss Thi...
 
Thermal Engineering -unit - III & IV.ppt
Thermal Engineering -unit - III & IV.pptThermal Engineering -unit - III & IV.ppt
Thermal Engineering -unit - III & IV.ppt
 
Booking open Available Pune Call Girls Pargaon 6297143586 Call Hot Indian Gi...
Booking open Available Pune Call Girls Pargaon  6297143586 Call Hot Indian Gi...Booking open Available Pune Call Girls Pargaon  6297143586 Call Hot Indian Gi...
Booking open Available Pune Call Girls Pargaon 6297143586 Call Hot Indian Gi...
 
Online banking management system project.pdf
Online banking management system project.pdfOnline banking management system project.pdf
Online banking management system project.pdf
 
Intze Overhead Water Tank Design by Working Stress - IS Method.pdf
Intze Overhead Water Tank  Design by Working Stress - IS Method.pdfIntze Overhead Water Tank  Design by Working Stress - IS Method.pdf
Intze Overhead Water Tank Design by Working Stress - IS Method.pdf
 
Double Revolving field theory-how the rotor develops torque
Double Revolving field theory-how the rotor develops torqueDouble Revolving field theory-how the rotor develops torque
Double Revolving field theory-how the rotor develops torque
 
Roadmap to Membership of RICS - Pathways and Routes
Roadmap to Membership of RICS - Pathways and RoutesRoadmap to Membership of RICS - Pathways and Routes
Roadmap to Membership of RICS - Pathways and Routes
 
notes on Evolution Of Analytic Scalability.ppt
notes on Evolution Of Analytic Scalability.pptnotes on Evolution Of Analytic Scalability.ppt
notes on Evolution Of Analytic Scalability.ppt
 
ONLINE FOOD ORDER SYSTEM PROJECT REPORT.pdf
ONLINE FOOD ORDER SYSTEM PROJECT REPORT.pdfONLINE FOOD ORDER SYSTEM PROJECT REPORT.pdf
ONLINE FOOD ORDER SYSTEM PROJECT REPORT.pdf
 
CCS335 _ Neural Networks and Deep Learning Laboratory_Lab Complete Record
CCS335 _ Neural Networks and Deep Learning Laboratory_Lab Complete RecordCCS335 _ Neural Networks and Deep Learning Laboratory_Lab Complete Record
CCS335 _ Neural Networks and Deep Learning Laboratory_Lab Complete Record
 
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
 

When the AIs failures send us back to our own societal biases

  • 1.
  • 2. 2 The good and the evil
  • 3. 3 The good and the evil
  • 4. 4 Mélanie DUCOFFE PhD., Research Data Scientist @mducoffe melanie.ducoffe@airbus.com Deep Learning, AI explainability About me !
  • 5. 5 Clément DUFFAU PhD., Lead DevOps @clement0210 clement.duffau@stack-labs.com Automation, verification & validation, assurance quality, safety About me !
  • 6. 6 Summary Artificial Intelligence and ethics Data selection biases In production biases Train interpretation biases
  • 7. 7 Job posts gathered into a single position: prone to many biases. Data biases and Model biases What do you think of combining : architect + dev+ testing + maintenance ? The nature of AI into the work market Data Scientist Data Analyst Data Engineer Machine Learning Scientist Machine Learning Engineer Statistician Find, clean and organize data for companies Transform and manipulate large datasets to suit the analysis of companies Perform batch processing or real time processing on gathered/stored data Research new data approaches and algorithms Create data funnels and deliver software solutions Interpret, analyze and report statistical information
  • 8. 8 ● Bias (in statistics) : “the difference between the expectation of the sample estimator and the true population value, which reduces the representativeness of the estimator by systematically distorting it” ● “The big takeaway is that we don’t know what we don’t know,” (Alice Popejoy) What do you mean about biases ?
  • 11. 11 ● MIT study decision-making of self-driving car on killing scenario ● Scenario with ○ high/low level of education ○ young/old ○ male/female ○ pets ○ traffic signal respect http://moralmachine.mit.edu/ The Moral Machine
  • 12. 12 ● Results “rank” you on ○ number of death ○ law respect ○ gender,age,health, ... ● Need ethical choices at the government, insurance ? manufacturer ? passengers ? The Moral Machine
  • 13. 13 ● 3 different clusters ● Cultural differences ○ USA/Europe kill the oldest vs Japon saves the oldest ○ Colombia saves high educated people vs Finland doesn’t matter ○ South america/France saves women The Moral Machine experiment, Edmond Awad et al. (2018), Nature The Moral Machine
  • 14. 14 Ethics in clinical investigations ● Technical committee to review scientific foundations and safety (in France, ANSM) ● Ethics committee on animal AND humans (in France, CPP) ○ Composed by medical professional and citizen ○ Review application form on ■ benefit/risk ■ information quality ■ resources to conduct the study ■ patient recruitment process ■ Patient consent modality ○ Follow the experiment process
  • 15. 15 Applying ethics committee in AI at Google ● Advanced Technology External Advisory Council (ATEAC) ○ Ensure the white paper principles for AI at Google ■ Be socially beneficial ■ Avoid creating or reinforcing unfair bias ■ Be built and tested for safety ■ Be accountable to people ■ Incorporate privacy design principles ■ Uphold high standards of scientific excellence ■ Be made available for uses that accord with these principles ● Dissolved 1 week after creation ○ Ethics of some attendees were discussed … ○ Questioning themselves of the needs to represent every part of the society
  • 17. 17 ● In the 90s, “I was breaking all of [my classmates’] facial-recognition software because apparently all the pictures they were taking were of people with significantly less melanin than I have” (Charles Isbell) ● In 2015, the “gorilla mistake” in Google Photos ● 25 years, not the same learning model at all, but the same root cause Face recognition from the 90s to nowadays
  • 18. 18 ● In 2014, Amazon develops an IA to “find” key success factor and hire these people ● Train on their 10 last years hired people ● 89% of the engineering workforce is male ● At the end, you have sexist recruitment AI ! ● Representative data inside the company by not inside the society … AI recruitment by Amazon
  • 19. 19 Biases in the Features Outcome outputFeatures Model Diploma Gender Hobbies Nationality University ... HIRED ? Model Outcome output Features ● Lack of an appropriate set of features ● Lack of an appropriate dataset ● Imbalanced dataset or bias in the output ● Unawareness: remove sensitive features from your data
  • 20. 20 Tutorial: your first bias detector Survival of passengers on the Titanic ● Decision Tree ● Leaves = class labels ● Nodes = splitting/conjunctions of features ● Important nodes = less deep, lot of observations ● Many Framework coexist: FairML, “What If”, IBM Bias Assessment ● ‘No universal solution’: combine them Summarize: Your chance of survival were good iff: - you were a woman - you were a young boy with few siblings
  • 21. 21 ● Bad advice pointed out by IBM in internal in 2017 ○ ex : Suggest a cancer patient with severe bleeding be given a drug that could cause the bleeding to worsen ● Started off by using real patient data ● Fed it with hypothetical data ● “Synthetic cases allow you to treat and train Watson on a variety of patient variables and conditions that might not be present in random patient samples, but are important to treatment recommendations” (Edward Barbani) ● Pointed out the difficulties to collect representative data in medical Unsafe medical recommendations by IBM Watson
  • 22. 22 Data Augmentation ● Historically for images: rotation, flipping, adding noise… ● Object detection models : performance loss on corrupted images ● CNNs generalize poorly to novel distortion types, despite being trained on a variety of other distortions
  • 23. 23 Generating new data with GANS source: Medium, “Generative Adversarial with Maths” by Madhu Sanjeevi Discriminator Generator Generator
  • 24. 24 Understanding biases and Generalization in Deep Generative Models Can we learn the distribution of data with GANS ? What are the inductive biases of Deep Generative models ? ● Unbiased and consistent density estimation impossible ● Inductive biases ● Similar cognitive bias as humans: numerosity ● Weber’s law: relative change (ratio) True data Generated data
  • 25. 25 GAN-based Data Augmentation Perpetuates biases
  • 26. 26
  • 28. 28 Group Fairness ● Demographic Parity ○ ex : gender independent ● Equal odds ○ ex : take into account the reality statistics in one side ● Equal opportunity ○ ex : take into account the reality statistics in both side
  • 31. 31 Black box models: non linear decision functions DOG CAT WHY
  • 32. 32 LIME: Zooming ! create neighbors and deduce a linear classifier DOG CAT DOG CAT
  • 33. 33 Is it working ? Examples of Explanations are usually cherry-picked !
  • 34. 34 Individual Fairness for Explainability The fact that 2 resulting saliency maps are different is fundamentally due to the network itself being fragile to such perturbations
  • 37. 37 ● Experienced lane recognition algorithm ● Doesn’t expect that 3 stickers break the robustness ● Hacking the AI with thinks that doesn’t matter humans Tesla autonomous driving hacked with stickers
  • 39. 39 Hacking tutorial: design your own adversarial example CAT DOG CAT DOG if B is predicted as a cat, retry to cross the decision boundary with B ! Linear Model NON Linear Model A A B B
  • 40. 40 Hacking Tutorial: I don’t have access to the model This is not a car ! This is not a car !
  • 41. 41 Defending against adversarial attacks ?! SUCCESS FAILURES 2013 Discovery of adversarial examples 2015 Fast Gradient Sign 2015 DeepFool 2016 Carlini Wagner 2019 Unforeseen Attack 2013 Adversarial Training BRUTE FORCE: training with adv 2015 Defensive Distillation output probabilities rather than hard decisions 2016 JPEG compression against adv
  • 42. 42 Why is it hard to defend against adversarial examples? ● Adaptative: block one type of attacks but leaves vulnerability open to an attacker who knows the defense being used ● Hard to defend because hard to understand the theory ?
  • 43. 43 Adversarial Examples are not Bugs, they are features BIRD DOG CAT “Bird” “Dog” “Cat” create the label-target adversarial dataset DOG CAT BIRD “Dog” “Cat” “Bird”
  • 45. 45 Conclusion Data Biases Model Biases Explanation Prod Biases Adv Examples If no one cares, it is highly likely that the next person who suffers from biased treatment is one of us
  • 46. 46 What about in the future ? ● Art project exposed racial bias in the biggest image dataset ImageNet in Sept. 2019 ● ImageNet will remove 600,000 images ● Number of publications per year with the name Imagenet in the abstract…