Conformal Prediction for Federated Uncertainty Quantification Under Label Shift

Authors

Vincent Plassier, Lagrange Mathematics and Computing Research Center
Mehdi Makni, Lagrange Mathematics and Computing Research Center
Aleksandr Rubashevskii, Skolkovo Institute of Science and Technology
Eric Moulines, École Polytechnique & Mohamed bin Zayed University of Artificial IntelligenceFollow
Maxim Panov, Technology Innovation Institute

Document Type

Conference Proceeding

Publication Title

Proceedings of Machine Learning Research

Abstract

Federated Learning (FL) is a machine learning framework where many clients collaboratively train models while keeping the training data decentralized. Despite recent advances in FL, the uncertainty quantification topic (UQ) remains partially addressed. Among UQ methods, conformal prediction (CP) approaches provides distribution-free guarantees under minimal assumptions. We develop a new federated conformal prediction method based on quantile regression and take into account privacy constraints. This method takes advantage of importance weighting to effectively address the label shift between agents and provides theoretical guarantees for both valid coverage of the prediction sets and differential privacy. Extensive experimental studies demonstrate that this method outperforms current competitors.

First Page

27907

Last Page

27947

Publication Date

7-2023

Keywords

Conformal predictions, Decentralised, Distribution-free, Learning frameworks, Machine-learning, Prediction methods, Quantile regression, Train model, Training data, Uncertainty quantifications

Comments

Open Access version from PMLR

Uploaded on June 21, 2024

Recommended Citation

V. Plassier, M. Makni, A. Rubashevskii, E. Moulines and M. Panov, "Conformal Prediction for Federated Uncertainty Quantification Under Label Shift," Proceedings of Machine Learning Research, vol. 202, pp. 27907 - 27947, Jul 2023.