FAIRNESS & EXPLAINABILITY in MACHINE LEARNING (Spring 2023)
Instructor: Chih-Duo Hong
Lecture Hours: Wed D56
Lecture Room: 260305
Office Hours: By appointment
Contact: chihduo@nccu.edu.tw
This course will introduce selected topics in machine learning fairness, explainability, and safety, with a focus on approaches providing provable correctness and quality guarantee. The course will consist of three parts: lectures, group discussions, and paper presentations. In the lectures, selected research topics and results will be introduced in a self-contained manner. During the group discussion sessions, students will discuss in groups over selected scientific articles and case studies. For the paper presentation sessions, students will read and present recent technical papers. After taking this course, students will gain a general knowledge of Trustworthy AI, as well as a deep understanding of specific techniques for practicing and researching formal AI fairness, explainability, and safety.
Lectures
Week 01. Course overview
Week 02. Fairness metrics (I)
Week 03. Fairness metrics (II)
Week 04. Counterfactual fairness
Week 05. Abductive and contrastive explanations (I)
Week 06. Abductive and contrastive explanations (II)
Week 07. Logical approaches to XAI (I)
Week 09. Logical approaches to XAI (II)
Week 10. Property inference
Week 11. Anchor explanations
Week 12. Counterfactual explanations (I)
Week 16. Counterfactual explanations (II)
Week 17. Adversarial robustness (I)
Week 18. Adversarial robustness (II)
Presentation Paper List
From Contrastive to Abductive Explanations and Back Again
https://alexeyignatiev.github.io/assets/pdf/inams-aiia20-preprint.pdf
On Relating Explanations and Adversarial Examples
https://alexeyignatiev.github.io/assets/pdf/inms-nips19-preprint.pdf
Which Neural Network Makes More Explainable Decisions?
https://link.springer.com/article/10.1007/s10515-022-00338-w
Towards Formal Approximated Minimal Explanations of Neural Networks
https://arxiv.org/pdf/2210.13915.pdf
Finding Common Ground for Incoherent Horn Expressions
https://arxiv.org/pdf/2209.06455.pdf
DL2: Training and Querying Neural Networks with Logic
http://proceedings.mlr.press/v97/fischer19a/fischer19a.pdf
On the Explanatory Power of Boolean Decision Trees
https://arxiv.org/abs/2108.05266
Counterfactual Explanations without Opening the Black Box
https://arxiv.org/abs/1711.00399
Constraint-Driven Explanations for Black Box ML Models
https://ojs.aaai.org/index.php/AAAI/article/view/20805
Formalizing the Robustness of Counterfactual Explanations for Neural Networks
https://arxiv.org/pdf/2208.14878
Learning to Deceive with Attention-Based Explanations
https://arxiv.org/pdf/1909.07913
A Unifying and General Account of Fairness Measurement in Recommender Systems
https://www.sciencedirect.com/science/article/pii/S0306457322002163
DeepRED: Rule Extraction from Deep Neural Networks
https://link.springer.com/chapter/10.1007/978-3-319-46307-0_29
Formal Security Analysis of Neural Networks using Symbolic Intervals
https://arxiv.org/pdf/1804.10829
Using MaxSAT for Efficient Explanations of Tree Ensembles
https://alexeyignatiev.github.io/assets/pdf/iisms-aaai22-preprint.pdf
Explaining and Interpreting LSTMs
https://arxiv.org/pdf/1909.12114
A Moral Framework for Understanding of Fair ML
https://arxiv.org/pdf/1809.03400
Inherent Trade-Offs in the Fair Determination of Risk Scores
https://arxiv.org/abs/1609.05807
Strategic Classification is Causal Modeling in Disguise
http://proceedings.mlr.press/v119/miller20b/miller20b.pdf
On Efficiently Explaining Graph-Based Classifiers
https://arxiv.org/pdf/2106.01350
Globally-Robust Neural Networks
https://arxiv.org/pdf/2102.08452
On Computing Probabilistic Explanations for Decision Trees
https://arxiv.org/pdf/2207.12213
The Bouncer Problem: Challenges to Remote Explainability
https://arxiv.org/abs/1910.01432
ART: Abstraction Refinement-Guided Training for Neural Networks
https://arxiv.org/abs/1907.10662
Identifying and Correcting Label Bias in Machine Learning
https://arxiv.org/abs/1901.04966
Discussion Articles
Formalizing Fairness
https://cacm.acm.org/magazines/2022/8/262911-formalizing-fairness
Explainable AI: Opening the Black Box or Pandora’s Box?
https://cacm.acm.org/magazines/2022/4/259398-explainable-ai
Want a job? You'll Have to Convince Our AI Bot First
https://www.cbc.ca/news/business/recruitment-ai-tools-risk-bias-hidden-workers
Great Promise but Potential for Peril
https://news.harvard.edu/.../ethical-concerns-mount-as-ai-takes-bigger-decision-making/
Scientists Increasingly Can’t Explain How AI Works
https://www.vice.com/en/.../scientists-increasingly-cant-explain-how-ai-works
How Big Data is Unfair
https://medium.com/@mrtz/how-big-data-is-unfair-9aa544d739de
Can AI’s Recommendations Be Less Insidious?
https://spectrum.ieee.org/recommendation-engine-insidious
Are You Still Using Real Data to Train Your AI?
https://spectrum.ieee.org/synthetic-data-ai
Racial Bias Found in Algorithms That Determine Health Care for Millions of Patients
https://spectrum.ieee.org/racial-bias-found-in-algorithms-that-determine-health-care
Engineering Bias Out of AI
https://spectrum.ieee.org/engineering-bias-out-of-ai
Moving Beyond "Algorithmic Bias Is a Data Problem"
https://www.cell.com/action/showPdf?pii=S2666-3899%2821%2900061-1
Are We Witnessing the Dawn of Post-theory Science?
https://www.theguardian.com/.../are-we-witnessing-the-dawn-of-post-theory-science
Machine Bias: Risky Assessment in Criminal Sentencing
https://www.propublica.org/article/machine-bias-risk-assessments-in-criminal-sentencing
Amazon Scraps Secret AI Recruiting Tool That Showed Bias Against Women
https://www.reuters.com/article/us-amazon-com-jobs-automation-insight-idUSKCN1MK08G
What Happens When an Algorithm Cuts Your Health Care
https://www.theverge.com/.../healthcare-medicaid-algorithm-arkansas-cerebral-palsy
The Parable of Google Flu
https://gking.harvard.edu/files/gking/files/0314policyforumff.pdf
It’s Too Easy to Hide Bias in Deep-Learning Systems
https://spectrum.ieee.org/its-too-easy-to-hide-bias-in-deeplearning-systems
Conservatives Are Panicking About AI Bias, Think ChatGPT Has Gone 'Woke'
https://www.vice.com/.../conservatives-panicking-about-ai-bias
Crime-Prediction Tool May Be Reinforcing Discriminatory Policing
https://www.businessinsider.com/predictive-policing-discriminatory-police-crime
There’s an easy way to make lending fairer for women. Trouble is, it’s illegal.
https://www.technologyreview.com/.../theres-an-easy-way-to-make-lending-fairer-for-women/
The scary truth about AI copyright is nobody knows what will happen next
https://www.theverge.com/.../generative-ai-copyright-infringement-legal-fair-use-training-data
Why It’s So Damn Hard to Make AI Fair and Unbiased
https://www.vox.com/future-perfect/22916602/ai-bias-fairness-tradeoffs-artificial-intelligence
Artificial Disinformation: Can Chatbots Destroy Trust on the Internet?
https://medium.com/.../artificial-disinformation-can-chatbots-destroy-trust-on-the-internet
AI Doesn’t Have to Be This Way
https://spectrum.ieee.org/ai-skeptics
Protecting AI Models from “Data Poisoning”
https://spectrum.ieee.org/ai-cybersecurity-data-poisoning
Preventing AI From Divulging Its Own Secrets
https://spectrum.ieee.org/how-prevent-ai-power-usage-secrets
Further Reading
Following is a list of surveys and articles about AI explanability, safety, and fairness.
Explainable AI
NIST: Four Principles of Explainable Artificial Intelligence, 2021
NIST: Psychological Foundations of Explainability and Interpretability in Artificial Intelligence, 2021
Amina Adadi and Mohammed Berrada: Peeking Inside the Black-Box: A Survey on Explainable Artificial Intelligence (XAI), 2018
Tim Miller: Explanation in Artificial Intelligence: Insights from the Social Sciences, 2018
Plamen P. Angelov et al.: Explainable Artificial Intelligence: An Analytical Review, 2021
Miruna A. Clinciu and Helen F. Hastie: A Survey of Explainable AI Terminology, 2019
Roberto Confalonieri et al.: A Historical Perspective of Explainable Artificial Intelligence, 2020
Derek Doran et al.: What Does Explainable AI Really Mean? A New Conceptualization of Perspectives, 2017
Doshi-Velez and Kim: Towards a Rigorous Science of Interpretable Machine Learning, 2017
Gabriëlle Ras et al.: Explanation Methods in Deep Learning: Users, Values, Concerns and Challenges, 2018
Safe AI
Xiaowei Huang et al.: A Survey of Safety and Trustworthiness of Deep Neural Networks: Verification, Testing, Adversarial Attack and Defense, and Interpretability, 2020
Sina Mohseni, et al.: Taxonomy of Machine Learning Safety: A Survey and Primer, 2022
Dario Amodei et al.: Concrete Problems in AI Safety, 2016
Fair AI
Solon Barocas et at.: Fairness and Machine Learning, 2022
Benjaminvan Giffen et al.: Overcoming the Pitfalls and Perils of Algorithms, 2022
Zeyu Tang et al.: What-Is and How-To for Fairness in Machine Learning: A Survey, Reflection, and Perspective, 2022
EU’s Rights Agency: Bias in Algorithms - Artificial Intelligence and Discrimination, 2022
Ninareh Mehrabi et al.: A Survey on Bias and Fairness in Machine Learning, 2019
Alexandra Chouldechova and Aaron Roth: A Snapshot of the Frontiers of Fairness in Machine Learning, 2020
Sorelle Friedler et al.: The (Im)possibility of Fairness, 2021
Sam Corbett-Davies and Sharad Goel: The Measure and Mismeasure of Fairness: A Critical Review of Fair Machine Learning, 2018
Formal XAI
Caterina Urban and Antoine Mine: A Review of Formal Methods applied to Machine Learning, 2021
J. Marques-Silva: Logic-Based Explainability in Machine Learning, 2022
A. Darwiche: Three Modern Roles for Logic in AI, 2020
Opinions
Jasmijn Bastings and Katja Filippova: The elephant in the interpretability room: Why use attention as explanation when we have saliency methods? 2020
Alon Jacovi and Yoav Goldberg: Aligning Faithful Interpretations with their Social Attribution, 2021
Michael Kearns and Aaron Roth: The Ethical Algorithm: The Science of Socially Aware Algorithm Design, 2019
Randy Goebel et al.: Explainable AI: The New 42 ?, 2018
Zachary C. Lipton: The Mythos of Model Interpretability, 2016
Cathy O'Neill: Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy, 2016
Boyd and Crawford: Critical Questions for Big Data, 2012
Frank Pasquale: The Black Box Society, 2015