Summary of Federated Split Learning with Model Pruning and Gradient Quantization in Wireless Networks, by Junhe Zhang et al.
Federated Split Learning with Model Pruning and Gradient Quantization in Wireless Networks
by Junhe Zhang, Wanli Ni, Dongyu Wang
First submitted to arxiv on: 9 Dec 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Distributed, Parallel, and Cluster Computing (cs.DC); Networking and Internet Architecture (cs.NI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary Federated learning typically requires all edge devices to train a complete model locally, but as AI models increase in scale, limited resources on edge devices become a bottleneck for efficient fine-tuning. This paper proposes a lightweight federated split learning (FedSL) scheme that alleviates the training burden by pruning client-side models dynamically and using quantized gradient updates to reduce computation overhead. The proposed scheme also applies random dropout to activation values at the split layer to reduce communication overhead. Simulation results verify the effectiveness and advantages of the proposed lightweight FedSL in wireless network environments, demonstrating improved convergence performance. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Federated learning helps many devices work together to make AI models. But as these models get bigger, they need more power to train on small devices like smartphones. This paper has a new way to help with this problem. They divide the model into smaller parts and have each device work on one part at a time. This makes it faster and uses less power. The paper also shows how to make it even better by getting rid of some parts that are not important and using special math tricks to save time and power. |
Keywords
* Artificial intelligence * Dropout * Federated learning * Fine tuning * Pruning
