Research map / Federated learning and privacy / Privacy attacks and secure learning
Securing Secure Aggregation: Mitigating Multi-Round Privacy Leakage in Federated Learning
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use batch-structured Multi-RoundSecAgg instead of random selection | Federated learning user selection · IID; CIFAR-100, CIFAR-10, and MNIST | long-term privacy guarantee: T=3,4,6 vs T=1 | Helped |
| Use batch-structured Multi-RoundSecAgg instead of random selection | Federated learning user selection · CIFAR-100 non-IID | test accuracy: 44.32% -> 47.06% | Helped |
| Use batch-structured Multi-RoundSecAgg instead of random selection | Federated learning user selection · CIFAR-10 non-IID | test accuracy: 45.20% -> 47.21% | Helped |
| Use batch-structured Multi-RoundSecAgg instead of random selection | Federated learning user selection · MNIST non-IID | test accuracy: 85.79% -> 94.16% | Helped |
| Use batch-structured Multi-RoundSecAgg instead of random selection | Federated learning user selection · IID settings across CIFAR-100, CIFAR-10, and MNIST | test accuracy: 49.15% vs 50.22%; 64.64% vs 64.80%; 98.21% vs 98.15% | No clear effect |
| Increase privacy guarantee from T=3 to T=6 (ablation) | Multi-RoundSecAgg · CIFAR-100 IID | test accuracy: 50.22% -> 42.89% | Hurt |
| Increase privacy guarantee from T=3 to T=6 (ablation) | Multi-RoundSecAgg · CIFAR-100 non-IID | test accuracy: 47.06% -> 39.57% | Hurt |
| Increase privacy guarantee from T=3 to T=6 (ablation) | Multi-RoundSecAgg · CIFAR-10 IID | test accuracy: 64.80% -> 65.01% | No clear effect |
| Increase privacy guarantee from T=3 to T=6 (ablation) | Multi-RoundSecAgg · CIFAR-10 non-IID | test accuracy: 47.21% -> 46.35% | Hurt |
| Increase privacy guarantee from T=3 to T=6 (ablation) | Multi-RoundSecAgg · MNIST IID | test accuracy: 98.15% -> 97.72% | No clear effect |
| Increase privacy guarantee from T=3 to T=6 (ablation) | Multi-RoundSecAgg · MNIST non-IID | test accuracy: 94.16% -> 89.88% | Hurt |
| Use user partitioning rather than Multi-RoundSecAgg with T=3 | Federated learning user selection · CIFAR-100 IID | test accuracy: 50.22% vs 25.73% | Hurt |
Closest papers by meaning
- Janus: Dual-Server Multi-Round Secure Aggregation with Verifiability for Federated Learning (ICML 2025 · similarity 0.91)
- Towards Trustworthy Federated Learning with Untrusted Participants (ICML 2025 · similarity 0.90)
- Noise-Aware Algorithm for Heterogeneous Differentially Private Federated Learning (ICML 2024 · similarity 0.89)
- FedAA: A Reinforcement Learning Perspective on Adaptive Aggregation for Fair and Robust Federated Learning (AAAI 2025 · similarity 0.89)
- Self-Driven Entropy Aggregation for Byzantine-Robust Heterogeneous Federated Learning (ICML 2024 · similarity 0.88)
- Dual Defense: Enhancing Privacy and Mitigating Poisoning Attacks in Federated Learning (NeurIPS 2024 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.