BriarPatches: Pixel-Space Interventions for Inducing Demographic Parity
This provides a user-level intervention mechanism for fairness in AI, addressing a gap where prior methods were only accessible to experts.
The paper tackles the problem of inducing demographic parity in pre-trained classifiers by introducing BriarPatches, pixel-space interventions that obscure sensitive attributes from model representations, resulting in downstream predictors that exhibit demographic parity without requiring model developer expertise.
We introduce the BriarPatch, a pixel-space intervention that obscures sensitive attributes from representations encoded in pre-trained classifiers. The patches encourage internal model representations not to encode sensitive information, which has the effect of pushing downstream predictors towards exhibiting demographic parity with respect to the sensitive information. The net result is that these BriarPatches provide an intervention mechanism available at user level, and complements prior research on fair representations that were previously only applicable by model developers and ML experts.