FairCal: Fairness Calibration for Face Verification
This addresses fairness issues in face verification, particularly for minority groups impacted by law enforcement use, offering a practical solution without common drawbacks like accuracy loss.
The paper tackled bias in face recognition models by introducing FairCal, a post-training method that increased accuracy, reduced false positive rate gaps across ethnicities, and achieved state-of-the-art results without requiring sensitive attribute knowledge or retraining.
Despite being widely used, face recognition models suffer from bias: the probability of a false positive (incorrect face match) strongly depends on sensitive attributes such as the ethnicity of the face. As a result, these models can disproportionately and negatively impact minority groups, particularly when used by law enforcement. The majority of bias reduction methods have several drawbacks: they use an end-to-end retraining approach, may not be feasible due to privacy issues, and often reduce accuracy. An alternative approach is post-processing methods that build fairer decision classifiers using the features of pre-trained models, thus avoiding the cost of retraining. However, they still have drawbacks: they reduce accuracy (AGENDA, PASS, FTC), or require retuning for different false positive rates (FSN). In this work, we introduce the Fairness Calibration (FairCal) method, a post-training approach that simultaneously: (i) increases model accuracy (improving the state-of-the-art), (ii) produces fairly-calibrated probabilities, (iii) significantly reduces the gap in the false positive rates, (iv) does not require knowledge of the sensitive attribute, and (v) does not require retraining, training an additional model, or retuning. We apply it to the task of Face Verification, and obtain state-of-the-art results with all the above advantages.