Predictive Multiplicity in Classification

3 code implementations • ICML 2020 • Charles T. Marx, Flavio du Pin Calmon, Berk Ustun

We apply our tools to measure predictive multiplicity in recidivism prediction problems.

Classification General Classification +1

Disentangling Influence: Using Disentangled Representations to Audit Model Predictions

1 code implementation • NeurIPS 2019 • Charles T. Marx, Richard Lanas Phillips, Sorelle A. Friedler, Carlos Scheidegger, Suresh Venkatasubramanian

Specifically, we show that disentangled representations provide a mechanism to identify proxy features in the dataset, while allowing an explicit computation of feature influence on either individual outcomes or aggregate-level outcomes.