Summary of Beyond Local Sharpness: Communication-efficient Global Sharpness-aware Minimization For Federated Learning, by Debora Caldarola et al.
Beyond Local Sharpness: Communication-Efficient Global Sharpness-aware Minimization for Federated Learning
by Debora Caldarola, Pietro Cagnasso, Barbara Caputo, Marco Ciccone
First submitted to arxiv on: 4 Dec 2024
Categories
- Main: Artificial Intelligence (cs.AI)
- Secondary: None
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 This paper proposes a novel federated learning approach called FedGloSS (Federated Global Server-side Sharpness) that prioritizes optimizing global sharpness on the server, using client-side sharpness-aware minimization (SAM). This approach addresses the issue of data heterogeneity across edge devices by ensuring models converge to flatter minima, resulting in better generalization and robustness. The method cleverly approximates sharpness using previous global gradients, eliminating the need for additional client communication. Evaluations show that FedGloSS outperforms state-of-the-art FL methods on various federated vision benchmarks. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper helps us build a better way to train models with lots of devices sharing their data. Right now, it’s hard because each device has different information, so the model doesn’t work well everywhere. The new approach, called FedGloSS, makes sure the model is good for all devices by optimizing its sharpness on the main server. It does this without needing more communication from the devices, which saves time and energy. This means we can get better results when training models that combine data from lots of different places. |
Keywords
» Artificial intelligence » Federated learning » Generalization » Sam
