| Date | Lecture | Required Reading | Reading Assignments | Project |
|
Apr 8 | Lecture 1 Biases and Fairness
slides, video | Barocas: Ch 2
Bishop: Ch 8.2 | - On Formalizing Fairness in Prediction with Machine Learning, ArXiv 2017
- k-NN as an implementation of situation testing for discrimination discovery and prevention, SIGKDD 2011
- Equality of Opportunity in Supervised Learning, NeurIPS 2016
- Discrimination-aware Data Mining, SIGKDD 2008
- Fairness Constraints: Mechanisms for Fair Classification, AISTATS 2017
| Project Guideline Out |
| Apr 10 | Lecture 2 Fair Representation Learning
slides, video | |
- Data decisions and theoretical implications when adversarially learning fair representations, FAT 2017
- Inherent trade-offs in the fair determination of risk scores, ArXiv 2016
- Fairgan: Fairness-aware generative adversarial networks. IEEE Big Data 2018
- Flexibly fair representation learning by disentanglement, ICML 2019
- Wasserstein Fair Classification. UAI 2019
| |
| Apr 15 | Lecture 3 Interpretability and Transparency
slides, video | Molnar: Ch 2, Ch 4 | - The mythos of model interpretability. Queue 2018
- Peeking inside the black-box: A survey on Explainable Artificial Intelligence (XAI), IEEE Access 2018
- Towards a rigorous science of interpretable machine learning. Arxiv 2017
- Falling rule lists. AISTATS 2015
- Discovering interpretable representations for both deep generative and discriminative models, ICML 2018
| |
| Apr 17 | Lecture 4 Post Hoc Interpretability and Proxy Models
slides, video | Molnar: Ch 5.7, Ch 5.8 | - "Why should i trust you?" Explaining the predictions of any classifier, SIGKDD 2016
- Interpretable decision sets: A joint framework for description and prediction, SIGKDD 2016
- Distilling knowledge from deep networks with applications to healthcare domain, Arxiv 2015
- Model agnostic supervised local explanations, NeurIPS 2018
- Explaining classifications for individual instances. IEEE Transactions on Knowledge and Data Engineering 2008
| |
| Apr 22 | Lecture 5 Feature Interaction for interpretability
slides, video | Molnar: Ch 5.9, Ch 5.10 | - Towards Efficient Data Valuation Based on the Shapley Value, ICML 2019
- A Game Theoretic Approach to Class-wise Selective Rationalization, NeurIPS 2019
- Granger-causal attentive mixtures of experts: Learning important features with neural networks, AAAI 2019
- GNNExplainer: Generating explanations for graph neural networks, NeurIPS 2019
- Ancona, Marco, Cengiz Öztireli, and Markus Gross. Explaining deep neural networks with a polynomial time algorithm for shapley values approximation, ICML 2019
| |
|
Apr 24 | Lecture 6 Example and Visualization Based Methods for Interpretability
slides, video | Molnar: Ch 6 | - Counterfactual explanations without opening the black box: Automated decisions and the GDPR, Harv. JL & Tech, 2017
- INVASE: Instance-wise variable selection using neural networks, ICLR 2018
- On Concept-Based Explanations in Deep Neural Networks, arXiv 2019
- Efficient Data Representation by Selecting Prototypes with Importance Weights, ICDM 2019
- The bayesian case model: A generative approach for case-based reasoning and prototype classification. NeurIPS 2014
| Project Proposal Due |
|
Apr 29 | Lecture 7 Interpreting deep neural networks
slides, video | |
- Learning how to explain neural networks: PatternNet and PatternAttribution, ICLR 2018
- Towards robust interpretability with self-explaining neural networks, NeurIPS 2018
- Visualizing and understanding convolutional networks, ECCV 2014
- Real time image saliency for black box classifiers, NeurIPS 2017
- Sanity checks for saliency maps, NeurIPS 2018
| |
|
May 1 | Lecture 8 Fairness Through Input Manipulation
slides, video | |
- From parity to preference-based notions of fairness in classification, NeurIPS 2017
- A reductions approach to fair classification, ICML 2018
- On fairness and calibration, NeurIPS 2017
- Predict responsibly: improving fairness and accuracy by learning to defer, NeurIPS 2018
- A common framework to provide explanations and analyse the fairness and robustness of black-box models, AIES 2020
| |
| May 6 | Lecture 9 Fair NLP
slides, video | |
- Lipstick on a Pig: Debiasing Methods Cover up Systematic Gender Biases in Word Embeddings But do not Remove Them, NAACL 2019
- Gender Bias in Contextualized Word Embeddings, NAACL 2019
- Understanding the Origins of Bias in Word Embeddings, ICML 2019
- The Woman Worked as a Babysitter: On Biases in Language Generation, EMNLP 2019
- The risk of racial bias in hate speech detection, ACL 2019
| |
| May 8 | Lecture 10 Fairness for Vision Representations
slides, video | |
- Counterfactual fairness, NeurIPS 2017
- Men also like shopping: Reducing gender bias amplification using corpus-level constraints, EMNLP 2017
- Feature transfer learning for face recognition with under-represented data, CVPR 2019
- Fairness of exposure in rankings, KDD 2017
- Gender shades: Intersectional accuracy disparities in commercial gender classification, FAccT 2018
| |
| May 13 | Lecture 11 Mid-way Presentation | | | Midterm Report Due |
| May 15 | Lecture 12 Robustness and Adversarial Attacks
slides, video1, video2 | |
- Darts: Deceiving autonomous cars with toxic signs. arXiv 2018
- Adversarial examples are not bugs, they are features, NeurIPS 2019
- Synthesizing robust adversarial examples, ICML 2018
- Universal adversarial perturbations, CVPR 2017
- Deepfool: a simple and accurate method to fool deep neural networks, CVPR 2016
| |
|
May 20 | Lecture 13 Adversarial Defense
slides, video | | - On detecting adversarial perturbations, ICLR 2017
- Certified defenses against adversarial examples, ICLR 2018
- Certified adversarial robustness via randomized smoothing, ICML 2019
- Defense-gan: Protecting classifiers against adversarial attacks using generative models, ICLR 2018
- Ensemble adversarial training: Attacks and defenses, ICLR 2018
| |
|
May 22 | Lecture 14 Fair Causal Reasoning
slides, video | | - PC-Fairness: A Unified Framework for Measuring Causality-based Fairness, NeurIPS 2019
- Path-specific counterfactual fairness, AAAI 2019
- Envy-free classification, NeurIPS 2019
- Causal inference for social discrimination reasoning, Journal of Intelligent Information Systems 2019
- Equality of opportunity in classification: A causal approach, NeurIPS 2018
| |
|
May 27 | Lecture 15 (cancaled) | | - Conditional learning of fair representations, ICLR 2020
- Inherent tradeoffs in learning fair representations, NeurIPS 2019
- A Geometric Solution to Fair Representations, AAAI/ACM AI, Ethics, and Society 2020
- Learning Certified Individually Fair Representations, arXiv 2020
- A general approach to fairness with optimal transport, AAAI 2020
| |
| May 29 | Lecture 16 Disentangled Fair Representation and ML Auditing
slides, video | | - Concrete problems in AI safety, arXiv 2016
- Inherent trade-offs in the fair determination of risk scores, arXiv 2016
- The concept of fairness in the GDPR: a linguistic and contextual interpretation, FAccT 2020
- European Union regulations on algorithmic decision-making and a “right to explanation”, AI magazine 2017
- AI Fairness 360: An extensible toolkit for detecting, understanding, and mitigating unwanted algorithmic bias, arXiv 2018
| |
| Jun 3 | Lecture 17 Privacy
slides, video | | - Practical secure aggregation for privacy-preserving machine learning, ACM SIGSAC Conference on Computer and Communications Security, 2017
- Federated learning: Strategies for improving communication efficiency, arXiv 2016
- Towards federated learning at scale: System design, SysML 2019
- Federated multi-task learning, NeurIPS 2017
- Calibrating noise to sensitivity in private data analysis, Theory of cryptography conference 2006
| |
| Jun 5 | Final Project Presentation | | | |
| Jun 10 | Final Project Presentation | | | |
| Jun 12 | | | | 11:59PM PT: Final Report Due
|