Michael P. Kim

mpkim@

about

I am a Miller Postdoctoral Fellow at UC Berkeley, hosted by Shafi Goldwasser. Prior to this, I completed my Ph.D. in the Stanford Theory Group under the sage guidance of Omer Reingold.

My research investigates foundational questions about responsible machine learning. Much of my work aims to identify ways in which machine-learned predictors can exhibit problematic behavior (e.g., unfair discrimination) and develop algorithmic tools that provably mitigate such behaviors. More broadly, I am interested in how the computational lens (i.e., algorithms and complexity theory) can provide insight into emerging societal and scientific challenges.

selected publications & recent manuscripts [all]

Backward Baselines: Is your model predicting the past? [arXiv]
Moritz Hardt and MPK
preprint 2022

Planting Undetectable Backdoors in Machine Learning Models [arXiv]
Shafi Goldwasser, MPK, Vinod Vaikuntanathan, Or Zamir
preprint 2022

Low-Degree Multicalibration [arXiv]
Parikshit Gopalan, MPK, Mihir Singhal, Shengjia Zhao
COLT 2022

Universal Adaptability: Target-Independent Inference that Competes with Propensity Scoring
MPK, Christoph Kern, Shafi Goldwasser, Frauke Kreuter, Omer Reingold
PNAS 2022

Beyond Bernoulli: Generating Random Outcomes that cannot be Distinguished from Nature
Cynthia Dwork, MPK, Omer Reingold, Guy N. Rothblum, Gal Yona
ALT 2022

Outcome Indistinguishability [arXiv] [ECCC]
Cynthia Dwork, MPK, Omer Reingold, Guy N. Rothblum, Gal Yona
STOC 2021

Multiaccuracy: Black-Box Post-Processing for Fairness in Classification [arXiv]
MPK, Amirata Ghorbani, James Zou
AAAI AI, Ethics, and Society 2019

Calibration for the (Computationally-Identifiable) Masses [arXiv]
Úrsula Hébert-Johnson, MPK, Omer Reingold, Guy N. Rothblum
ICML 2018