We will have discussion sessions throughout the semester. Typically, you will read several papers and provide a short response in Markdown to be uploaded to the class.
(1/29) Discussion 1.
Remember to sign up for leading small-group discussion and written assignment for one of the below:
Track how technology is transforming work, Tom Mitchell & Erik Brynjolfsson and Mitchell slides on work and Mitchell video on work
Preparing for the future of AI, NSTC
Artificial Intellence, automation and the economy, executive office of the president
Beyond prediction: Using big data for policy problems. Athey
Plan to lead discussion of your set of readings for 5-10 minutes. Possible questions for discussion:
- What are the new directions of machine learning?
- How are academic and policymakers addressing the real-world impacts of machine learning?
- In what ways should machine learning be regulated? If so, how, and if not, why?
- How can methods be adapted for varying quality of data?
- What should the machine learning community pay more attention to?
Discussion 2. Fairness 2/28
Sign up on Canvas for a reading: you may choose from one of the below or another source (the spreadsheet gives you a few places to start from). For class, prepare a 3-slide presentation including an overview, a mathematical formulation and example, and a critique. You will each give informal 3-5 minute mini presentations during class followed by a discussion. For the slides a template is available on Canvas.
Fairness Beyond Disparate Treatment and Disparate Impact: Learning Classification without Disparate Mistreatment. Zafar et al.
Certifying and Removing Disparate Impact. Feldman et al.
A multidisciplinary survey on discrimination analysis
Automated decision systems used by agences (New York City Council)
Fairness in ML tutorial
Machine Learning and Prediction in Medicine -- Beyond the Peak of Inflated Expectations. Chen and Asch
Unintended Consequences of Machine Learning in Medicine. Cabitza et al.
Big Data, Machine Learning, and Clinical Medicine. Obermeyer and Emanuel
Machine Learning and the Profession of Medicine. Darcy et al.
Is Medicine Mesmerized by Machine Learning?
Discussion c. Temporal modeling
Discussion d. Causality
Learning Representations for Counterfactual Inference
CausalGAN: Learning Causal Implicit Generative Models with Adversarial Training
Reliable Decision Support using Counterfactual Models
Estimation and Inference of Heterogeneous Treatment Effects using Random Forests?
Synth-Validation: Selecting the Best Causal Inference Method for a Given Dataset