Federated learning communication time cost predicted for wireless networks

Researchers have developed a method to predict the communication time cost in federated learning scenarios over IEEE 802.11 wireless networks. By using ns-3 simulations to measure frame delivery ratios and saturation throughput under various client densities and loads, they created a proxy for update admission probability. This proxy, combined with an equation, estimates communication time, showing that while round counts to reach target accuracy remained stable, communication time increased significantly with client density. A Bianchi-anchored estimator achieved a mean absolute percentage error between 2.3% and 10.2% on unseen configurations, though this error was measured against the same estimation equation rather than independent completion time. AI

IMPACT Provides a method to estimate communication overhead in federated learning over wireless networks, potentially aiding in the optimization of distributed AI training.

RANK_REASON Academic paper detailing a novel method for predicting communication costs in federated learning over wireless networks. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• alphaXiv
• arXiv
• Bianchi
• CatalyzeX
• DagsHub
• FedAvg
• federated learning
• Gotit.pub
• Hugging Face
• IArxiv
• IEEE 802.11
• ns-3
• ScienceCast

• paper
• infra

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →