Research

 fairness in machine learning, online learning, bandit problems, computational learning theory, differential privacy, combinatorics. 

Papers

[4] Accuracy First: Selecting a Differential Privacy Level for Accuracy Constrained ERM. [NIPS 2017]

[3] A Convex Framework for Fair Regression. Submitted, May 2017.

[2] A Framework for Meritocratic Fairness of Online Linear Models. Submitted, May 2017.

Fairness in Linear Bandit Problems   Fairness, Accountability, and Transparency in Machine Learning (FATML) 2016. Presented at NYU Law School, November 2016. Full version: Rawlsian Fairness for Machine Learning.  with: Matt Joseph, Michael Kearns, Jamie Morgenstern, Aaron Roth

[1]  Aztec Castles and the dP3 Quiver

[The Journal of Physics A: Mathematical and Theoretical, 2015]

Mahalanobis Matching and Equal Percent Bias Reduction

Senior Honors Thesis Harvard University 2015, High Honors. Supervised by Natesh Pillai.

Plane Partitions and Domino TilingsIntel STS Semifinalist 2011.

A new bijection between domino tilings of aztec diamonds and plane partitions is developed, leading to a simple proof of the generating function. A purely combinatorial proof of the Aztec Diamond Theorem due to [EKLP] is given. PDF available upon request.

Advertisements