Preserving Privacy in Federated Learning with Ensemble Cross-Domain Knowledge Distillation

TL;DR

This paper introduces a privacy-preserving, communication-efficient federated learning method using ensemble cross-domain knowledge distillation with quantization and noise, achieving better accuracy and efficiency.

Contribution

It presents a novel one-shot offline knowledge distillation approach that enhances privacy and reduces communication in federated learning.

Findings

Outperforms baseline FL algorithms in accuracy

Reduces communication overhead significantly

Provides stronger privacy guarantees

Abstract

Federated Learning (FL) is a machine learning paradigm where local nodes collaboratively train a central model while the training data remains decentralized. Existing FL methods typically share model parameters or employ co-distillation to address the issue of unbalanced data distribution. However, they suffer from communication bottlenecks. More importantly, they risk privacy leakage. In this work, we develop a privacy preserving and communication efficient method in a FL framework with one-shot offline knowledge distillation using unlabeled, cross-domain public data. We propose a quantized and noisy ensemble of local predictions from completely trained local models for stronger privacy guarantees without sacrificing accuracy. Based on extensive experiments on image classification and text classification tasks, we show that our privacy-preserving method outperforms baseline FL algorithms with superior performance in both accuracy and communication efficiency.