Research map / Federated learning and privacy / Communication-efficient FL
HO-SFL: Hybrid-Order Split Federated Learning with Backprop-Free Clients and Dimension-Free Aggregation
Qiyuan Chen, Xian Wu, Yi Wang, Xianhao Chen
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 |
|---|---|---|---|
| replace client-side BP with hybrid-order ZO using server activation gradients | SFL · CIFAR-10 IID | validation accuracy: marginal gap to SFL | No clear effect |
| use HO-SFL instead of SFL in Non-IID setting | SFL · CIFAR-10 Non-IID | validation accuracy | Improved |
| add server-supplied activation gradients to client ZO | ZO-SFL · GLUE (LLaMA-3.2-1B) SST2 | accuracy (%): 61.5 -> 93.9 | Improved |
| hybrid-order instead of first-order SplitLoRA | SplitLoRA · GLUE RTE (LLaMA-3.2-1B) | accuracy (%): 70.0 -> 73.3 | Improved |
| hybrid-order instead of first-order SplitLoRA | SplitLoRA · GLUE WSC (OPT-125M) | accuracy (%): 64.4 -> 60.6 | Hurt |
| scale model from 125M to 8B | HO-SFL · SQuAD | F1: 0.9413 vs 0.9389 for Qwen3-8B | No clear effect |
| dimension-free scalar aggregation instead of model upload | SFL · CIFAR-10 IID | communication cost (GB): 4.20 -> 2.44 | Improved |
| eliminate client-side BP | SFL · LLaMA-3.2-1B SST-2 LoRA | client peak memory (GB): 8.78 -> 4.14 | Improved |
| increase client-side layers from 2 to 8 (ablation) | HO-SFL · LLaMA-3.2-1B | convergence speed: marginal reduction | Hurt |
| increase perturbation count P (ablation) | HO-SFL · ResNet-18 | performance: monotonic improvement | Improved |
| choose smoothing parameter mu too large or too small (ablation) | HO-SFL · ablation | approximation quality | Mixed |
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.
- FSL-SAGE — FSL-SAGE: Accelerating Federated Split Learning via Smashed Activation Gradient Estimation (ICML 2025) · better on 1 of 1 shared result
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
- Convergence Analysis of Split Federated Learning on Heterogeneous Data (NeurIPS 2024 · similarity 0.90)
- Towards Straggler-Resilient Split Federated Learning: An Unbalanced Update Approach (NeurIPS 2025 · similarity 0.88)
- FSL-SAGE: Accelerating Federated Split Learning via Smashed Activation Gradient Estimation (ICML 2025 · similarity 0.87)
- Achieving Dimension-Free Communication in Federated Learning via Zeroth-Order Optimization (ICLR 2025 · similarity 0.87)
- MultiSFL: Towards Accurate Split Federated Learning via Multi-Model Aggregation and Knowledge Replay (AAAI 2025 · similarity 0.87)
- FedQS: Optimizing Gradient and Model Aggregation for Semi-Asynchronous Federated Learning (NeurIPS 2025 · similarity 0.86)
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.
