Project

Throughout the semester, students will work on an open-ended project, investigating material related to one of the recent course topics. The aim of the project is to explore the primary literature on the topic in more depth than covered in lectures. Successful course projects may serve as a starting point for research that eventually results in a conference publication.

Projects will be evaluated based on multiple components, with the following key dates.

Suggested Topics/Papers

• Calibration and Multicalibration
  • Testing Calibration in Nearly-Linear Time [Hu-Jambulapati-Tian-Yang]
  • On Computationally Efficient Multi-Class Calibration [Gopalan-Hu-Rothblum]
  • Loss Minimization Yields Multicalibration for Large Neural Networks [Blasiok-Gopalan-Hu-Kalai-Nakkiran]
  • When is Multicalibration Post-Processing Necessary? [Hansen-Devic-Nakkiran-Sharan]
  • When Does Optimizing a Proper Loss Yield Calibration? [Blasiok-Gopalan-Hu-Nakkiran]
  • A Unifying Theory of Distance from Calibration [Blasiok-Gopalan-Hu-Nakkiran]
  • From Pseudorandomness to Multi-Group Fairness and Back [Dwork-Lee-Lin-Tankala]
  • Complexity-Theoretic Implications of Multicalibration [Casacuberta-Dwork-Vadhan]
  • Multicalibration for Confidence Scoring in LLMs [Detommaso-Bertran-Fogliato-Roth]
• Omniprediction
  • Omnipredictors for regression and the approximate rank of convex functions [Gopalan-Okoroafor-Raghavendra-Shetty-Singhal]
  • U-Calibration: Forecasting for an Unknown Agent [Kleinberg-Paes Leme-Schneider-Teng]
  • Optimal Multiclass U-Calibration Error and Beyond [Luo-Senapati-Sharan]
  • Oracle Efficient Online Multicalibration and Omniprediction [Garg-Jung-Reingold-Roth]
  • Forecasting for Swap Regret for All Downstream Agents [Roth-Shi]
• Multi-Group-X
  • Multigroup Robustness [Hu-Peale-Shen]
  • Comparative Learning: A Sample Complexity Theory for Two Hypothesis Classes [Hu-Peale]
  • Learning from Outcomes: Evidence-Based Rankings [Dwork-Kim-Reingold-Rothblum-Yona]
  • Beyond Bernoulli: Generating Random Outcomes that cannot be Distinguished from Nature [Dwork-Kim-Reingold-Rothblum-Yona]
  • Stability and Multigroup Fairness in Ranking with Uncertain Predictions [Devic-Korolova-Kempe-Sharan]
  • The Statistical Scope of Multicalibration [Noarov-Roth]
  • Multi-Group Proportional Representation [Oesterling et al.]
  • Batch Multivalid Conformal Prediction [Jung-Noarov-Ramalingam-Roth]
• Performative Prediction / Validity
  • Who Leads and Who Follows in Strategic Classification? [Zrnic-Mazumdar-Sastry-Jordan]
  • Performative Power [Hardt-Jagadeesan-Mendler Dünner]
  • The Relative Value of Prediction in Algorithmic Decision Making [Perdomo]
  • Is your model predicting the past? [Hardt-Kim]
  • Causal Fairness for Outcome Control [Plecko-Bareinboim]
  • Causal Fairness Analysis [Plecko-Bareinboim]
• Verification of Learning
  • PAC Verification of Statistical Algorithms [Mutreja-Shafer]
  • Classical Verification of Quantum Learning [Caro-Hinsche-Ioannou-Nietner-Sweke]
  • Certifying Private Probabilistic Mechanisms [Bell-Goldwasser-Kim-Watson]
  • Models That Prove Their Own Correctness [Amit-Goldwasser-Paradise-Rothblum]
• Fairness and Beyond
  • Fairness in Matching under Uncertainty [Devic-Kempe-Sharan-Korolova]
  • Preference-Informed Fairness [Kim-Korolova-Rothblum-Yona]
  • On Fairness and Stability in Two-Sided Matchings [Karni-Rothblum-Yona]
  • Equilibria, Efficiency, and Inequality in Network Formation for Hiring and Opportunity [Dwork-Hays-Kleinberg-Raghavan]
  • Human Expertise in Algorithmic Prediction [Alur-Raghavan-Shah]
  • Oracle Efficient Algorithms for Groupwise Regret [Acharya-Arunachaleswaran-Kannan-Roth-Ziani]