Optimal Fair Aggregation of Crowdsourced Noisy Labels using Demographic Parity Constraints

TL;DR

This paper investigates fairness in crowdsourced noisy label aggregation, proposing methods to ensure demographic parity and providing theoretical guarantees for fairness convergence in the context of majority voting and Bayesian aggregation.

Contribution

It introduces a formal fairness analysis framework for crowdsourced label aggregation and extends a fairness post-processing algorithm to discrete labels, with proven convergence guarantees.

Findings

Fairness gap of Majority Vote is bounded by individual annotator fairness gaps.

Aggregated fairness converges exponentially fast to ground-truth fairness under certain conditions.

The proposed fairness enforcement method is effective on synthetic and real datasets.

Abstract

As acquiring reliable ground-truth labels is usually costly, or infeasible, crowdsourcing and aggregation of noisy human annotations is the typical resort. Aggregating subjective labels, though, may amplify individual biases, particularly regarding sensitive features, raising fairness concerns. Nonetheless, fairness in crowdsourced aggregation remains largely unexplored, with no existing convergence guarantees and only limited post-processing approaches for enforcing $\varepsilon$-fairness under demographic parity. We address this gap by analyzing the fairness s of crowdsourced aggregation methods within the $\varepsilon$-fairness framework, for Majority Vote and Optimal Bayesian aggregation. In the small-crowd regime, we derive an upper bound on the fairness gap of Majority Vote in terms of the fairness gaps of the individual annotators. We further show that the fairness gap of the aggregated consensus converges exponentially fast to that of the ground-truth under interpretable conditions. Since ground-truth itself may still be unfair, we generalize a state-of-the-art multiclass fairness post-processing algorithm from the continuous to the discrete setting, which enforces strict demographic parity constraints to any aggregation rule. Experiments on synthetic and real datasets demonstrate the effectiveness of our approach and corroborate the theoretical insights.