Research map / Federated learning and privacy / Privacy attacks and secure learning
Federated Spectral Clustering via Secure Similarity Reconstruction
Dong Qiao, Chris Ding, Jicong Fan
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 |
|---|---|---|---|
| Use FedSC (federated) instead of DSC | FedSC · COIL20 | clustering accuracy: 0.1009 ± 0.0100 -> 0.7828 ± 0.0231 | Improved |
| Use FedSC instead of vanilla SC on Bank | FedSC · Bank | clustering accuracy: 0.5918 ± 0.0000 -> 0.7672 ± 0.1457 | Improved |
| Use FedSC instead of vanilla SC on COIL20 | FedSC · COIL20 | clustering accuracy: 0.8025 ± 0.0009 -> 0.7828 ± 0.0231 | No clear effect |
| Use FedSC instead of vanilla SC on ORL | FedSC · ORL | clustering accuracy: 0.7865 ± 0.0106 -> 0.7235 ± 0.0170 | Hurt |
| Use FedSC instead of Kmeans on COIL20 | FedSC · COIL20 | clustering accuracy: 0.6113 ± 0.0534 -> 0.7828 ± 0.0231 | Improved |
| Use FedSC instead of vanilla SC on USPS under noise 0.3 sigma | FedSC · USPS | clustering accuracy: 0.6720 ± 0.0044 -> 0.6884 ± 0.0509 | Improved |
| Increase Gaussian noise from raw to 0.7 sigma on Iris (ablation) | FedSC · Iris | clustering accuracy: 0.9000 ± 0.0031 -> 0.6120 ± 0.0455 | Hurt |
| Split data across 8 clients instead of 1 (ablation) | FedSC · ORL | clustering accuracy: 0.7937 ± 0.0084 -> 0.7112 ± 0.0258 | Hurt |
| Split data across 8 clients instead of 1 (ablation) | FedSC · Iris | clustering accuracy: 0.9007 ± 0.0086 -> 0.8993 ± 0.0165 | No clear effect |
| Split data across 8 clients instead of 1 (ablation) | FedSC · Bank | clustering accuracy: 0.7536 ± 0.1271 -> 0.6821 ± 0.1405 | Hurt |
| Perturb factors C and Z (alpha_c 0 -> 0.15, alpha_z=0) (ablation) | FedSC · COIL20 | clustering accuracy: 0.7831±0.0268 -> 0.4835±0.0526 | Hurt |
| Perturb factor Z (alpha_z 0 -> 0.15, alpha_c=0) (ablation) | FedSC · COIL20 | clustering accuracy: 0.7831±0.0268 -> 0.7881±0.0223 | No clear effect |
Closest papers by meaning
- OmniFC: Rethinking Federated Clustering via Lossless and Secure Distance Reconstruction (NeurIPS 2025 · similarity 0.88)
- Subgraph Federated Learning via Spectral Methods (NeurIPS 2025 · similarity 0.88)
- Differentially Private Federated $k$-Means Clustering with Server-Side Data (ICML 2025 · similarity 0.87)
- Interaction-Aware Gaussian Weighting for Clustered Federated Learning (ICML 2025 · similarity 0.86)
- EBS-CFL: Efficient and Byzantine-robust Secure Clustered Federated Learning (AAAI 2025 · similarity 0.85)
- A Differentially Private Clustering Algorithm for Well-Clustered Graphs (ICLR 2024 · 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.
