Resilient Decentralized Wireless Federated Learning via Gradient Tracking with AdamW
What it is
This paper proposes QEF-GT-AdamW, a decentralized learning algorithm designed for wireless IoT edge networks. It combines gradient tracking for non-IID data, AdamW optimization, and dual-stream quantization with error feedback to reduce communication overhead and handle packet loss.
The problem it solves
Existing decentralized optimization schemes suffer from high communication overhead and degraded performance when operating under strict wireless constraints like limited bandwidth, fading channels, and packet losses.
What you could use it for
- Training machine learning models across distributed IoT devices with limited bandwidth
- Implementing decentralized learning in wireless networks with unreliable links and fading channels
Read the source ↗Back to the library →
Summary written by an AI model from the source; check the source before you rely on it.