[2603.05158] Balancing Privacy-Quality-Efficiency in Federated Learning through Round-Based Interleaving of Protection Techniques

Computer Science > Machine Learning arXiv:2603.05158 (cs) [Submitted on 5 Mar 2026] Title:Balancing Privacy-Quality-Efficiency in Federated Learning through Round-Based Interleaving of Protection Techniques Authors:Yenan Wang, Carla Fabiana Chiasserini, Elad Michael Schiller View a PDF of the paper titled Balancing Privacy-Quality-Efficiency in Federated Learning through Round-Based Interleaving of Protection Techniques, by Yenan Wang and 2 other authors View PDF Abstract:In federated learning (FL), balancing privacy protection, learning quality, and efficiency remains a challenge. Privacy protection mechanisms, such as Differential Privacy (DP), degrade learning quality, or, as in the case of Homomorphic Encryption (HE), incur substantial system overhead. To address this, we propose Alt-FL, a privacy-preserving FL framework that combines DP, HE, and synthetic data via a novel round-based interleaving strategy. Alt-FL introduces three new methods, Privacy Interleaving (PI), Synthetic Interleaving with DP (SI/DP), and Synthetic Interleaving with HE (SI/HE), that enable flexible quality-efficiency trade-offs while providing privacy protection. We systematically evaluate Alt-FL against representative reconstruction attacks, including Deep Leakage from Gradients, Inverting Gradients, When the Curious Abandon Honesty, and Robbing the Fed, using a LeNet-5 model on CIFAR-10 and Fashion-MNIST. To enable fair comparison between DP- and HE-based defenses, we introduce a new attacker...