Private and Distributionally Robust Federated Multiclass Classification
Abstract
Fault diagnosis in asset-intensive industries requires multiclass models trained on noisy, sensitive data collected across geographically distributed client sites. Federated learning enables collaborative training without centralizing data, but most approaches do not provide formal privacy guarantees for individual records stored at client sites and can degrade under label noise. To address these limitations, we develop a differentially private, distributionally robust framework for federated multiclass classification that provides sample-level privacy for records stored at each client. We build on a federated Wasserstein distributionally robust multiclass objective with the Crammer-Singer loss, solved using an Alternating Direction Method of Multipliers (ADMM)-based algorithm. To protect the released client updates, our method injects calibrated Gaussian noise into the proximal linearized client updates of both the local model and the local distributionally robust variable before aggregation at the central server, and we quantify the cumulative privacy loss using the moments accountant. We provide per-round sample-level (ε,δ)-differential privacy guarantees and an overall composed privacy guarantee across all training rounds, together with convergence results and an explicit privacy-utility tradeoff. Experiments on simulated industrial fault data and UCI benchmark datasets show that our method outperforms differentially private federated empirical risk minimization and stochastic gradient descent baselines in most tested settings, with the largest gains under strong privacy-induced noise.
Related
- Distributionally Robust Federated Learning with Multi-Source Data
- Data-driven joint optimization of maintenance and spare parts provisioning: A distributionally robust approach
- Conservative neural posterior estimation via distributionally robust training
- Personalized and differentially private federated learning for mobility data mining via the TrajectoryGuard framework
- Optimality of Affine Policies in Distributionally Robust Linear-Quadratic Control with Temporally Correlated Noise
BibTeX
@misc{hernandez2026private,
title = {{Private and Distributionally Robust Federated Multiclass Classification}},
author = {Andrea P. Hernandez and Michael Ibrahim and Nagi Gebraeel},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.13631},
eprint = {2610.13631},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.13631},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Private and Distributionally Robust Federated Multiclass Classification
AU - Andrea P. Hernandez
AU - Michael Ibrahim
AU - Nagi Gebraeel
PY - 2026
DA - 2026/10/09
PB - arXiv
DO - 10.48550/arxiv.2610.13631
UR - https://arxiv.org/abs/2610.13631
AB - Fault diagnosis in asset-intensive industries requires multiclass models trained on noisy, sensitive data collected across geographically distributed client sites. Federated learning enables collaborative training without centralizing data, but most approaches do not provide formal privacy guarantees for individual records stored at client sites and can degrade under label noise. To address these limitations, we develop a differentially private, distributionally robust framework for federated multiclass classification that provides sample-level privacy for records stored at each client. We build on a federated Wasserstein distributionally robust multiclass objective with the Crammer-Singer loss, solved using an Alternating Direction Method of Multipliers (ADMM)-based algorithm. To protect the released client updates, our method injects calibrated Gaussian noise into the proximal linearized client updates of both the local model and the local distributionally robust variable before aggregation at the central server, and we quantify the cumulative privacy loss using the moments accountant. We provide per-round sample-level (ε,δ)-differential privacy guarantees and an overall composed privacy guarantee across all training rounds, together with convergence results and an explicit privacy-utility tradeoff. Experiments on simulated industrial fault data and UCI benchmark datasets show that our method outperforms differentially private federated empirical risk minimization and stochastic gradient descent baselines in most tested settings, with the largest gains under strong privacy-induced noise.
ER -