Provable Optimization for Adversarial Fair Self-supervised Contrastive Learning
This work addresses fairness in self-supervised learning, which is incremental as it builds on existing adversarial fair representation learning methods by adding provable optimization guarantees.
The paper tackles the challenge of learning fair encoders in self-supervised learning with limited sensitive attribute annotations by proposing a stochastic algorithm, SoFCLR, that provably optimizes adversarial fair contrastive learning without requiring large batch sizes, achieving effectiveness in downstream classification across eight fairness notions.
This paper studies learning fair encoders in a self-supervised learning (SSL) setting, in which all data are unlabeled and only a small portion of them are annotated with sensitive attribute. Adversarial fair representation learning is well suited for this scenario by minimizing a contrastive loss over unlabeled data while maximizing an adversarial loss of predicting the sensitive attribute over the data with sensitive attribute. Nevertheless, optimizing adversarial fair representation learning presents significant challenges due to solving a non-convex non-concave minimax game. The complexity deepens when incorporating a global contrastive loss that contrasts each anchor data point against all other examples. A central question is ``{\it can we design a provable yet efficient algorithm for solving adversarial fair self-supervised contrastive learning}?'' Building on advanced optimization techniques, we propose a stochastic algorithm dubbed SoFCLR with a convergence analysis under reasonable conditions without requring a large batch size. We conduct extensive experiments to demonstrate the effectiveness of the proposed approach for downstream classification with eight fairness notions.