Research map / Federated learning and privacy / Communication-efficient FL
ProgFed: Effective, Communication, and Computation Efficient Federated Learning by Progressive Training
Hui-Po Wang, Sebastian Stich, Yang He, Mario Fritz
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 |
|---|---|---|---|
| progressive training instead of end-to-end training in federated learning | FedAvg, ResNet-18 · CIFAR-100 federated | accuracy (%): 52.08 ± 0.44 -> 53.23 ± 0.09 | Improved |
| progressive training instead of end-to-end training | FedAvg, ConvNet · EMNIST federated | accuracy (%): 85.75 ± 0.11 -> 85.67 ± 0.06 | No clear effect |
| progressive training in centralized training | ResNet/VGG · CIFAR-100 centralized | accuracy (%): 76.08±0.12 -> 75.84±0.28 (ResNet-18); 77.77±0.38 -> 78.57±0.33 (ResNet-152) | No clear effect |
| symmetric growth (outer to inner) instead of asymmetric growth in U-net (ablation) | ProgFed 3D-Unet · BraTS federated | communication cost change: -5.02% -> -63.60 % | Improved |
| combine ProgFed with FedAdam | ProgFed · CIFAR-100 federated | accuracy (%): 56.21 -> 60.55 | Improved |
| combine ProgFed with FedProx | ProgFed · EMNIST federated | accuracy (%): 86.36 -> 86.08 | No clear effect |
| combine ProgFed with quantization (LQ-2) | ResNet-18 federated · CIFAR-100 federated | accuracy (%): 47.26 -> 52.87 | Improved |
| add warm-up for newly added layers (ablation) | ProgFed · CIFAR-100 federated ResNet-18 | accuracy (%): 51.09 -> 53.23 | Improved |
| progressive training instead of layerwise training (ablation) | ProgFed · CIFAR-100 centralized ResNet-18 | accuracy (%): 72.40±0.16 -> 75.84±0.28 | Improved |
| progressive growth instead of random sub-model selection (ablation) | ProgFed · CIFAR-100 centralized ResNet-18 | accuracy (%): 74.38±0.97 -> 75.84±0.28 | Improved |
| progressive growth instead of partial update with final-head supervision (ablation) | ProgFed · CIFAR-100 centralized ResNet-18 | accuracy (%): 74.70±0.04 -> 75.84±0.28 | Improved |
| combine local and final-head supervision (Mixed) vs local-head only (ablation) | ProgFed · CIFAR-100 centralized ResNet-18 | accuracy (%): 75.84±0.28 -> 75.04±1.26 | Hurt |
Later papers that compared with this method
1 later accepted paper reports ProgFed as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Developmental Federated Tuning: A Cognitive-Inspired Paradigm for Efficient LLM Adaptation (ICLR 2026 · better on 3 of 3 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
- FedQS: Optimizing Gradient and Model Aggregation for Semi-Asynchronous Federated Learning (NeurIPS 2025 · similarity 0.88)
- FedRTS: Federated Robust Pruning via Combinatorial Thompson Sampling (NeurIPS 2025 · similarity 0.87)
- Revisiting Ensembling in One-Shot Federated Learning (NeurIPS 2024 · similarity 0.87)
- SparsyFed: Sparse Adaptive Federated Learning (ICLR 2025 · similarity 0.87)
- SpaFL: Communication-Efficient Federated Learning With Sparse Models And Low Computational Overhead (NeurIPS 2024 · similarity 0.86)
- Achieving Dimension-Free Communication in Federated Learning via Zeroth-Order Optimization (ICLR 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.
