Research map / Federated learning and privacy / Communication-efficient FL
SFedHIFI: Fire Rate-Based Heterogeneous Information Fusion for Spiking Federated Learning
Ran Tao, Qiugang Zhan, Shantian Yang, Xiurui Xie, Qi Tian, Guisong Liu
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 HIFI fusion module (ablation) | SFedHIFI · CIFAR-10 Dir(0.3) | avg top-1 accuracy: 78.16 -> 84.73 | Improved |
| add HIFI fusion module (ablation) | SFedHIFI · CIFAR-10 IID | avg top-1 accuracy: 88.71 -> 89.62 | Improved |
| select fusion layer by firing rate instead of random layer (ablation) | SFedHIFI · CIFAR-10 Dir(0.3) | avg top-1 accuracy: 67.48 -> 84.73 | Improved |
| select fusion layer by firing rate instead of fixed last conv layer (ablation) | SFedHIFI · CIFAR-10 Dir(0.3) | avg top-1 accuracy: 79.45 -> 84.73 | Improved |
| select fusion layer by firing rate instead of fixed last conv layer (ablation) | SFedHIFI · CIFAR-10 IID | avg top-1 accuracy: 88.52 -> 89.62 | Improved |
| random-layer fusion vs no fusion (ablation) | SFedHIFI · CIFAR-10 Dir(0.3) | avg top-1 accuracy: 78.16 -> 67.48 | Hurt |
| cross-scale aggregation (SFedHIFI) vs same-scale baselines | SNN FL · CIFAR-100 Dir(0.5) | avg top-1 accuracy: 21.42 -> 33.00 | Improved |
| cross-scale aggregation (SFedHIFI) vs same-scale baselines | SNN FL · CIFAR-10 Dir(0.3) | avg top-1 accuracy: 74.49 -> 84.73 | Improved |
| cross-scale aggregation (SFedHIFI) vs same-scale baselines | SNN FL · CIFAR-100 IID | avg top-1 accuracy: 39.08 -> 39.20 | Improved |
| cross-scale aggregation (SFedHIFI) vs same-scale baselines | SNN FL · Fashion-MNIST Dir(0.3) | avg top-1 accuracy: 88.41 -> 88.43 | No clear effect |
| replace ANN with SNN (SFedHIFI) | SFedHIFI · Fashion-MNIST | avg energy (x10^-3 mJ): 90.917 -> 51.148 | Improved |
| replace ANN with SNN (SFedHIFI) | SFedHIFI · Fashion-MNIST Dir(0.3) | avg top-1 accuracy: 91.06 -> 88.43 | Hurt |
Closest papers by meaning
- Sparse Spiking Neural Network: Exploiting Heterogeneity in Timescales for Pruning Recurrent SNN (ICLR 2024 · similarity 0.85)
- Spiking Transformer with Experts Mixture (NeurIPS 2024 · similarity 0.84)
- Enhanced Self-Distillation Framework for Efficient Spiking Neural Network Training (NeurIPS 2025 · similarity 0.84)
- HetSyn: Versatile Timescale Integration in Spiking Neural Networks via Heterogeneous Synapses (NeurIPS 2025 · similarity 0.84)
- Improving the Sparse Structure Learning of Spiking Neural Networks from the View of Compression Efficiency (ICLR 2025 · similarity 0.84)
- SpikingSSMs: Learning Long Sequences with Sparse and Parallel Spiking State Space Models (AAAI 2025 · similarity 0.83)
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.
