Provable Optimization for Adversarial Fair Self-supervised Contrastive Learning

TL;DR

This paper introduces a provable, efficient stochastic algorithm for adversarial fair self-supervised contrastive learning, addressing optimization challenges in non-convex minimax problems with theoretical guarantees.

Contribution

The paper develops SoFCLR, a novel stochastic algorithm with convergence guarantees for adversarial fair SSL, advancing the theoretical understanding and practical optimization of fair representation learning.

Findings

Effective fairness-ensuring contrastive learning demonstrated

Convergence analysis supports algorithm's efficiency

Outperforms baselines on multiple fairness metrics

Abstract

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.