Research map / Federated learning and privacy / Communication-efficient FL
DISTREAL: Distributed Resource-Aware Learning in Heterogeneous Systems
Martin Rapp, Ramin Khalili, Kilian Pfeiffer, Jörg Henkel
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 Pareto-optimized, per-layer filter dropout instead of one shared dropout rate | DISTREAL · CIFAR-10 and CIFAR-100, heterogeneous resources across devices | convergence speed | Improved |
| Adjust the dropout vector locally before each mini-batch as resources change | DISTREAL · CIFAR-10 and CIFAR-100, heterogeneous resources across devices and time | convergence speed | Improved |
| Use a smaller network sized for the weakest device | Federated learning · CIFAR-10, heterogeneous resources across devices | early convergence speed | Improved |
| Use a smaller network sized for the weakest device | Federated learning · FEMNIST, heterogeneous resources across devices | convergence speed | Hurt |
| Change from a shared dropout rate to optimized per-layer dropout rates (ablation) | DISTREAL · DenseNet-40 design-space exploration | trade-off between MAC requirements and convergence speed | Improved |
| Change from fixed dropout throughout a round to resource-aware selection at each mini-batch (ablation) | DISTREAL · CIFAR-10 and CIFAR-100, time-varying resources | convergence speed | Improved |
| Increase the rate of resource changes during a round | Federated Dropout and HeteroFL · CIFAR-10 and CIFAR-100, heterogeneous resources across devices and time | convergence speed | Hurt |
| Increase the rate of resource changes during a round | DISTREAL · CIFAR-10 and CIFAR-100, heterogeneous resources across devices and time | convergence speed | No clear effect |
| Apply filter-based structured dropout to convolutional layers | Neural network training · DenseNet-40 on Raspberry Pi 4 | training time | Improved |
| Use a per-layer dropout vector with rates limited to [0,0.5] (ablation) | DISTREAL · Design-time search | final accuracy | Improved |
Closest papers by meaning
- WHALE-FL: Wireless and Heterogeneity Aware Latency Efficient Federated Learning over Mobile Devices via Adaptive Subnetwork Scheduling (AAAI 2025 · similarity 0.86)
- Reviving Stale Updates: Data-Free Knowledge Distillation for Asynchronous Federated Learning (NeurIPS 2026 · similarity 0.85)
- Efficient Federated Learning against Heterogeneous and Non-stationary Client Unavailability (NeurIPS 2024 · similarity 0.85)
- FedSPU: Personalized Federated Learning for Resource-Constrained Devices with Stochastic Parameter Update (AAAI 2025 · similarity 0.85)
- Fed-DFA: Federated Distillation for Heterogeneous Model Fusion Through the Adversarial Lens (AAAI 2025 · similarity 0.85)
- FedUV: Uniformity and Variance for Heterogeneous Federated Learning (CVPR 2024 · similarity 0.84)
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.
