Research map / Continual learning / Online and replay-based
RC-FCL: Combating Asynchronous Concept Drift in Federated Continual Learning via Retrospective Calibration
Hang Su, Yijun Mo, Zhiyu Zhang, Yankai Jiang, Bo Liu, Yichen Li, 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 drift detection (D) (ablation) | RC-FCL w/o D · CIFAR100 (IID) | accuracy: 31.3 -> 36.5 | Improved |
| add drift adaptation (A) (ablation) | RC-FCL w/o A · CIFAR100 (IID) | accuracy: 27.3 -> 36.5 | Improved |
| add both drift detection and adaptation (ablation) | RC-FCL w/o D+A · CIFAR100 (IID) | accuracy: 23.1 -> 36.5 | Improved |
| add drift detection (D) (ablation) | RC-FCL w/o D · Tiny-ImageNet (IID) | accuracy: 25.7 -> 28.5 | Improved |
| add drift adaptation (A) (ablation) | RC-FCL w/o A · Tiny-ImageNet (IID) | accuracy: 23.2 -> 28.5 | Improved |
| add both drift detection and adaptation (ablation) | RC-FCL w/o D+A · Tiny-ImageNet (IID) | accuracy: 20.7 -> 28.5 | Improved |
| use Sinkhorn distance rather than MMD, KL divergence or cosine similarity for drift detection (ablation) | RC-FCL · CIFAR100 and Tiny-ImageNet drift detection | F1 score | Improved |
| use RC-FCL instead of FedTA | FedTA · CIFAR100 (IID) | accuracy: 32.14 -> 35.23 | Improved |
| use RC-FCL instead of FedTA | FedTA · Tiny-ImageNet (Dir 0.1) | accuracy: 20.92 -> 22.62 | Improved |
| use RC-FCL instead of FedCIL | FedCIL · CIFAR100 rounds to best accuracy | communication rounds: 711 -> 767 | Hurt |
| use RC-FCL instead of ReFed | ReFed · Upload/download per round | communication (MB/round): 44.72 -> 56.11 | Hurt |
Closest papers by meaning
- Accurate Forgetting for Heterogeneous Federated Continual Learning (ICLR 2024 · similarity 0.91)
- HERO: A Heterogeneity-Aware Benchmark Library for Federated Continual Learning (NeurIPS 2026 · similarity 0.89)
- Traceable Federated Continual Learning (CVPR 2024 · similarity 0.88)
- Handling Spatial-Temporal Data Heterogeneity for Federated Continual Learning via Tail Anchor (CVPR 2025 · similarity 0.88)
- C$^2$Prompt: Class-aware Client Knowledge Interaction for Federated Continual Learning (NeurIPS 2025 · similarity 0.88)
- Federated Continual Learning Goes Online: Uncertainty-Aware Memory Management for Vision Tasks and Beyond (ICLR 2025 · 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.
