Algorithmic Fairness Research
Check back soon for some more exciting updates on the research I am doing!
Current Research
I’m currently conducting research in algorithmic fairness, with the Prof. Omer Reingold. I’m (preliminarily) investigating domain adaptation in the context of algorithmic fairness from an information theoretic point of view.
More Impossibilities between Calibration and Balance
With Amir Zur, I furthered some impossibility results between the two group fairness notions calibration and equalized odds. A (then recent) paper had proposed a way to reconcile the two notions by constructing a classifier that satisfies equalized odds using a calibrated predictor given a single threshold, seemingly violating the impossibility result. Using that paper as inspiration, we added two more necessary conditions to the original impossibility result to reconcile the contradiction.