Research map / Federated learning and privacy / Personalized and heterogeneous FL
Lightweight Federated Incremental Learning via Decoupled Replay
Xiuying Wang, Yichen Li, Hang Su, Gaozhuo Liu, Shiwei Li, Chuang Zhao, Jiangming Shi, Imran Razzak
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| add mixup to uploaded features (ablation) | Li-FIL · CIFAR100 (α=1.0) | A(f): 33.20 -> 35.61 | Improved |
| add mixup to uploaded features (ablation) | Li-FIL · CIFAR10 (α=1.0) | A(f): 39.61 -> 40.29 | Improved |
| add differential privacy Laplace noise (ablation) | Li-FIL · CIFAR10 (α=1.0) | A(f): 40.25 -> 40.29 | No clear effect |
| add differential privacy Laplace noise (ablation) | Li-FIL · CIFAR100 (α=1.0) | A(f): 35.49 -> 35.61 | No clear effect |
| add classifier replay loss on synthetic features (ablation) | Li-FIL · CIFAR100 (α=1.0) | A(f): 32.97 -> 35.61 | Improved |
| add MMD global feature alignment loss (ablation) | Li-FIL · CIFAR100 (α=1.0) | A(f): 33.09 -> 35.61 | Improved |
| add contrastive local-structure alignment loss (ablation) | Li-FIL · CIFAR100 (α=1.0) | A(f): 31.52 -> 35.61 | Improved |
| add contrastive local-structure alignment loss (ablation) | Li-FIL · CIFAR10 (α=1.0) | average accuracy: 60.92 -> 62.56 | Improved |
| apply mixup and DP together instead of raw features | uploaded features · CIFAR10 feature inversion attack | attack accuracy: 69.24 -> 29.3 | Improved |
| use mixup alone instead of raw features | uploaded features · CIFAR10 feature inversion attack | attack accuracy: 69.24 -> 55.7 | Improved |
| use feature replay with CVAE instead of buffer-based ReFed+ | federated incremental learning · CIFAR100 (α=1.0) | A(f): 32.15 -> 35.61 | Improved |
| replace image-level CVAE input with ResNet18 features | server generator · communication and model size | transmission cost / CVAE parameters: 3,072 -> 512 values; 3.8M -> 0.17M parameters | Improved |
Closest papers by meaning
- Towards Efficient Replay in Federated Incremental Learning (CVPR 2024 · similarity 0.91)
- Class-wise Balancing Data Replay for Federated Class-Incremental Learning (NeurIPS 2025 · similarity 0.88)
- Federated Class-Incremental Learning: A Hybrid Approach Using Latent Exemplars and Data-Free Techniques to Address Local and Global Forgetting (ICLR 2025 · similarity 0.87)
- CAN: Leveraging Clients As Navigators for Generative Replay in Federated Continual Learning (ICML 2025 · similarity 0.87)
- Accurate Forgetting for Heterogeneous Federated Continual Learning (ICLR 2024 · similarity 0.86)
- HERO: A Heterogeneity-Aware Benchmark Library for Federated Continual Learning (NeurIPS 2026 · similarity 0.85)
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.
