Research map / Federated learning and privacy / Personalized and heterogeneous FL
FedTopo: Topology-Informed Representation Alignment in Federated Learning Under Non-I.I.D. Conditions
Ke Hu, Liyao Xiang, Peng Tang, Weidong Qiu
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 topological alignment loss (TAL) (ablation) | FedTopo · CIFAR-10 Q-skew+K-skew (Setting A) | accuracy (%): 53.90 -> 67.05 | Improved |
| add topological alignment loss (TAL) (ablation) | FedTopo · FMNIST severe L-skew (Setting B) | accuracy (%): 58.87 -> 73.67 | Improved |
| add topological alignment loss (TAL) (ablation) | FedTopo · CIFAR-100 L-skew+N-skew (Setting C) | accuracy (%): 47.23 -> 63.60 | Improved |
| select ResNet-18 block layer2 via TGBS instead of conv1 (ablation) | FedTopo · ResNet-18 block selection | accuracy (%): 80.62 -> 88.35 | Improved |
| select ResNet-18 block layer2 instead of deep layer4 (ablation) | FedTopo · ResNet-18 block selection | accuracy (%): 85.35 -> 88.35 | Improved |
| select CNN conv1 via TGBS instead of conv2 (ablation) | FedTopo · CNN block selection | accuracy (%): 86.16 -> 87.52 | Improved |
| use Smooth-Topo alpha schedule instead of Warm-up (ablation) | FedTopo · CIFAR-10 Setting A, round 10 | accuracy (%): 73.52 -> 76.84 | Improved |
| use Smooth-Topo alpha schedule instead of Warm-up (ablation) | FedTopo · CIFAR-10 Setting B, round 10 | accuracy (%): 67.21 -> 71.31 | Improved |
| use Linear-Topo alpha schedule instead of Warm-up in early training (ablation) | FedTopo · CIFAR-10 Setting A, round 1 | accuracy (%): 65.13 -> 69.15 | Improved |
| use Linear-Topo rather than Smooth-Topo schedule (ablation) | FedTopo · CIFAR-10 Setting A, round 10 | accuracy (%): 76.84 -> 75.27 | Hurt |
| FedTopo versus second-best baseline across partitions | FedTopo · FMNIST, CIFAR-10, CIFAR-100 non-IID | accuracy (%): up to 8.69%, 11.05%, 8.44% | Improved |
| FedTopo versus FedDC on FMNIST L-skew | FedTopo · FMNIST L-skew | accuracy (%): 91.08 -> 91.30 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- FedSA — FedSA: A Unified Representation Learning via Semantic Anchors for Prototype-based Federated Learning (AAAI 2025) · better on 12 of 12 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- FedNE: Surrogate-Assisted Federated Neighbor Embedding for Dimensionality Reduction (NeurIPS 2024 · similarity 0.87)
- Less is More: Federated Graph Learning with Alleviating Topology Heterogeneity from A Causal Perspective (ICML 2025 · similarity 0.86)
- Beyond the Federation: Topology-aware Federated Learning for Generalization to Unseen Clients (ICML 2024 · similarity 0.86)
- FedSA: A Unified Representation Learning via Semantic Anchors for Prototype-based Federated Learning (AAAI 2025 · similarity 0.86)
- Topology-Aware Representation Alignment for Semi-Supervised Vision-Language Learning (NeurIPS 2026 · similarity 0.85)
- Personalized Subgraph Federated Learning with Differentiable Auxiliary Projections (NeurIPS 2025 · 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.
