A Stochastic Optimization Framework for Private and Fair Learning From Decentralized Data

Abstract: Machine learning models are often trained on sensitive data (e.g., medical records and race/gender) that is distributed across different “silos” (e.g., hospitals). These federated learning models may then be used to make consequential decisions, such as allocating health- care resources. Two key challenges emerge in this setting: (i) maintaining the privacy of each person’s data, even if other silos or an adversary with access to the central server tries to infer this data; (ii) ensuring that decisions are fair to different demographic groups (e.g., race/gender). In this paper, we develop a novel algorithm for private and fair federated learning (FL). Our algorithm satisfies inter-silo record-level differential privacy (ISRL-DP), a strong notion of private FL requiring that each silo’s communicated messages satisfy record-level differential privacy. In addition to being differentially private, our framework can be used to promote different fairness notions, including demographic parity and equalized odds. We prove that our algorithm converges under mild smoothness assumptions on the loss function (even in nonconvex settings), whereas prior work required strong convexity for convergence. As a byproduct of our analysis, we obtain the first convergent algorithm for ISRL-DP optimization of nonconvex-strongly concave min-max loss functions in federated learning. This convergent DP optimization algorithm is a valuable contribution in its own right. Additionally, our experiments demonstrate the state-of-the-art fairness-accuracy tradeoffs of our algorithm across different privacy levels. Compared to existing state of the art, we obtained an average of around 64% reduction in demographic parity fairness violation and 95% lower for equalized odds.

Submission Type: Regular submission (no more than 12 pages of main content)

Changes Since Last Submission: N/A

Assigned Action Editor: ~Audra_McMillan1

Submission Number: 7699