Research map / Federated learning and privacy / Differential privacy
Differentially Private Federated Learning With Local Regularization and Sparsification
Check 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 BLUR and LUS together instead of DP-FedAvg | DP-FedAvg · EMNIST, CNN-2-Layers, ε=2.0 | accuracy: 69.65±0.74 -> 74.48±0.52 | Helped |
| Use BLUR and LUS together instead of DP-FedAvg | DP-FedAvg · EMNIST, ResNet-18, ε=2.0 | accuracy: 73.52±0.53 -> 78.58±0.39 | Helped |
| Use BLUR and LUS instead of DP-FedAvg | DP-FedAvg · CIFAR-10, CNN-2-Layers, α=0.1 | accuracy: 53.84±1.04 -> 58.95±0.95 | Helped |
| Use BLUR and LUS instead of DP-FedAvg | DP-FedAvg · CIFAR-10, ResNet-18, α=100 | accuracy: 66.58±0.60 -> 69.42±0.47 | Helped |
| Add BLUR regularization (ablation) | DP-FedAvg · EMNIST, ResNet-18, ε=6.0 | test accuracy | Helped |
| Add LUS at sparsity 0.1 (ablation) | DP-FedAvg · EMNIST, ResNet-18, ε=6.0 | accuracy: 76.24 -> 76.52 | Helped |
| Add LUS at sparsity 0.5 (ablation) | DP-FedAvg · EMNIST, ResNet-18, ε=6.0 | accuracy: 76.24 -> 77.75 | Helped |
| Add LUS at sparsity 0.9 (ablation) | DP-FedAvg · EMNIST, ResNet-18, ε=6.0 | accuracy: 76.24 -> 77.39 | Helped |
| Add LUS with BLUR at sparsity 0.7 (ablation) | DP-FedAvg + BLUR · EMNIST, ResNet-18, ε=6.0 | accuracy: +4.08 | Helped |
| Increase sparsity from 0.7 to 0.9 with BLUR and LUS (ablation) | DP-FedAvg + BLUR + LUS · EMNIST, ResNet-18, ε=6.0 | accuracy: 80.32 -> 80.17 | Hurt |
| Add LUS at sparsity 0.5 alongside BLUR (ablation) | DP-FedAvg + BLUR · EMNIST, ResNet-18, ε=6.0 | accuracy: +3.73 | Helped |
| Use BLUR and LUS rather than DP-FedAvg as local training steps increase | DP-FedAvg · CIFAR-10, varying local update steps Q | accuracy | Helped |
Closest papers by meaning
- Private Model Personalization Revisited (ICML 2025 · similarity 0.90)
- Towards the Robustness of Differentially Private Federated Learning (AAAI 2024 · similarity 0.89)
- Noise-Aware Algorithm for Heterogeneous Differentially Private Federated Learning (ICML 2024 · similarity 0.89)
- Mitigating the Privacy–Utility Trade-off in Decentralized Federated Learning via f-Differential Privacy (NeurIPS 2025 · similarity 0.89)
- Privacy-Preserving Federated Convex Optimization: Balancing Partial-Participation and Efficiency via Noise Cancellation (ICML 2025 · similarity 0.89)
- Clients Collaborate: Flexible Differentially Private Federated Learning with Guaranteed Improvement of Utility-Privacy Trade-off (ICML 2025 · similarity 0.89)
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.