Privacy-enhanced federated learning via asynchronous aggregation and local differential perturbation
Federated learning framework combining dynamic differential privacy, homomorphic encryption, and local DP retains 82.6% accuracy at epsilon 0.1 while cutting communication 21.3%.
The paper proposes a privacy-enhanced federated learning framework integrating Dynamic Differential Privacy, lightweight Homomorphic Encryption, and Local Differential Privacy during training. An asynchronous aggregation strategy with version control supports distributed training in asynchronous environments. On CIFAR-10 and Purchase-100, the method maintains up to 82.6% classification accuracy under stringent privacy constraints (epsilon = 0.1) and reduces communication overhead by 21.3% versus FedAvg.
- Combines DDP, lightweight HE, and LDP in a federated training pipeline
- Asynchronous aggregation with version control handles asynchronous distributed settings
- 82.6% accuracy at epsilon 0.1; communication overhead cut 21.3% versus FedAvg
From the original article
1 min read at arxiv.orgThis study proposes a privacy-enhanced federated learning framework to address secure collaborative training in distributed data environments. The framework integrates Dynamic Differential Privacy (DDP), lightweight Homomorphic Encryption (HE), and Local Differential Privacy (LDP) mechanisms to ensure data privacy protection during model training.
By Zhen Zhong · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.