Responsible Machine Learning

Webpage for the course Responsible Machine Learning at PhD school QPE@UW

Meetings

We meet at 9 here: http://meet.drwhy.ai/

• 2021-03-02 Introduction to Responsible Machine Learning

• 2021-03-09 Project: topics selection

• 2021-03-16 If needed, let's talk a bit about tools, optional, no presentation planned

• 2021-03-23 Fairness - Legal perspective (Ethics) (Joanna R)

• 2021-03-30 Fairness - Introduction + Classification (Błażej P)

• 2021-04-13 Fairness - Testing discrimination (Anna B) (Krzysztof O) slides, jupiter notebook

• 2021-04-20 Interpretability - DARPA, 2 (Erita N) slides

• 2021-04-27 Project: first presentation

• 2021-05-04 Interpretability - Z Lipton (Konrad K)

• 2021-05-11 Interpretability - LIME (Stanislaw Ł)

• 2021-05-18 Interpretability - C Rudin (Sylwia W)

• 2021-05-25 Methods - Partial Dependence Plots (Pirapat P)

• 2021-06-01 Methods - Permutational Feature Importance (Mateusz B)

• 2021-06-08 Let's meet and conclude this semester

• 2021-06-15 Project: last presentation

Comments from meetings

• The Productivity–Pay Gap https://www.epi.org/productivity-pay-gap/
• The gap between wages and productivity https://blogs.lse.ac.uk/netuf/2018/05/24/the-gap-between-wages-and-productivity/
• ARTIFICIAL INTELLIGENCE AND THE MODERN PRODUCTIVITY PARADOX: A CLASH OF EXPECTATIONS AND STATISTICS https://www.nber.org/system/files/working_papers/w24001/w24001.pdf
• China: AI Governance Principles Released https://www.loc.gov/law/foreign-news/article/china-ai-governance-principles-released/

Projects

See https://pbiecek.github.io/xai_stories/

I expect you go though followng stepss

1. Find an small but interesting dataset. If in doubt: https://github.com/pbiecek/InterpretableMachineLearning2020/issues
2. Create an simple ML model
3. Test fairness
4. Try tools for explainability
5. Prepare report / presentation

Project groups

• Mateusz B. + Konrad K.
• Erita N. + Pirapat P. + Joanna R.
• Stanisław Ł. + Anna B.
• Sylwia W. + Błażej P.