Research map / Federated learning and privacy / Personalized and heterogeneous FL
Explainable Federated Learning via Global–Local Attribution Alignment
Dawood Wasif, Terrence J Moore, Chang-Tien Lu, Jin-Hee Cho
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 alignment loss (beta 0.2 vs 0) (ablation) | xFedAlign · MNIST | EDI: 0.0925 -> 0.0841 | Improved |
| add alignment loss (beta 0.2 vs 0) (ablation) | xFedAlign · MNIST | deletion/insertion AUC | No clear effect |
| increase alignment weight beta from 0 to 0.4 (ablation) | xFedAlign · MNIST | EDI: 0.0925 -> 0.0793 | Improved |
| add light DP noise (sigma 0.05 vs 0) (ablation) | xFedAlign · MNIST | EDI: 0.1330 -> 0.0819 | Improved |
| raise noise to sigma 0.2 (ablation) | xFedAlign · MNIST | EDI: 0.0819 -> 0.0852 | Hurt |
| increase top-k sparsity from 32 to 512 (ablation) | xFedAlign · MNIST | EDI: 0.0848 -> 0.0840 | No clear effect |
| increase surrogate epochs from 1 to 4 (ablation) | xFedAlign · MNIST | EDI: 0.0843 -> 0.1161 | Hurt |
| use xFedAlign instead of FedAvg-only training | federated learning · IMDb non-IID | accuracy: 0.550 -> 0.702 | Improved |
| use xFedAlign instead of FedAvg | federated learning · UCI Credit non-IID | accuracy: 0.315 -> 0.725 | Improved |
| use xFedAlign instead of FedAvg | federated learning · MNIST non-IID | accuracy: 0.930 -> 0.930 | No clear effect |
| use xFedAlign instead of Local-XAI | explanations · IMDb non-IID | EDI: 0.129 -> 0.043 | Improved |
| use interpretable-by-design Fed-XAI | federated model · MNIST non-IID | accuracy: 0.930 -> 0.449 | Hurt |
Closest papers by meaning
- FedXDS: Leveraging Model Attribution Methods to counteract Data Heterogeneity in Federated Learning (ICCV 2025 · similarity 0.90)
- GDFA: Geometry-Driven Federated Unlearning with Directional Task Vector Alignment (CVPR 2026 · similarity 0.87)
- Local or Global: Selective Knowledge Assimilation for Federated Learning with Limited Labels (ICCV 2023 · similarity 0.87)
- FLUX: Efficient Descriptor-Driven Clustered Federated Learning under Arbitrary Distribution Shifts (NeurIPS 2025 · similarity 0.87)
- LR-XFL: Logical Reasoning-Based Explainable Federated Learning (AAAI 2024 · similarity 0.86)
- Towards Diverse Device Heterogeneous Federated Learning via Task Arithmetic Knowledge Integration (NeurIPS 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.
