Authors: Xiaoxin Su, Yipeng Zhou, Laizhong Cui, Song Guo

Published on: February 06, 2024

Impact Score: 8.15

Arxiv code: Arxiv:2402.03815

Summary

• What is new: Introducing FediAC, an algorithm that utilizes programmable switch in federated learning for more efficient model update aggregation.
• Why this is important: The constraint of scarce memory space in programmable switches prevents efficient model aggregation in federated learning.
• What the research proposes: FediAC algorithm, which includes a client voting phase for determining significant model updates and a model aggregating phase that consumes less memory and communication traffic.
• Results: FediAC surpasses current state-of-the-art in model accuracy and reduces communication traffic.

Technical Details

Technological frameworks used: Federated Learning, Programmable Switch

Models used: FediAC algorithm

Data used: Public datasets

Potential Impact

Data privacy solutions, Cloud computing and Data Centers, Internet Service Providers

Want to implement this idea in a business?

We have generated a startup concept here: AgileNet.