Summary of Learning From Straggler Clients in Federated Learning, by Andrew Hard et al.
Learning from straggler clients in federated learning
by Andrew Hard, Antonious M. Girgis, Ehsan Amid, Sean Augenstein, Lara McConnaughey, Rajiv Mathews, Rohan Anil
First submitted to arxiv on: 14 Mar 2024
Categories
- Main: Machine Learning (cs.LG)
- 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 The paper investigates the effectiveness of existing federated learning algorithms when client devices return model updates with significant time delays. The authors develop Monte Carlo simulations to study synchronous optimization algorithms like FedAvg and FedAdam, as well as the asynchronous FedBuff algorithm, finding that these approaches struggle to learn from severely delayed clients. To improve upon this, they experiment with modifications and introduce two new algorithms, FARe-DUST and FeAST-on-MSG, based on distillation and averaging respectively. The authors demonstrate that their new algorithms outperform existing ones in terms of accuracy for straggler clients while providing better trade-offs between training time and total accuracy. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper looks at how well current machine learning algorithms work when devices send updates with a big delay. It simulates different scenarios to see what happens when devices take minutes, hours, or even days to send their updates. The researchers found that most of these algorithms don’t do well with delayed updates. To solve this problem, they tried some new approaches and came up with two new ways for devices to communicate: FARe-DUST and FeAST-on-MSG. They tested these new methods on a few different tasks and showed that they can do better than the old methods when dealing with slow devices. |
Keywords
* Artificial intelligence * Distillation * Federated learning * Machine learning * Optimization
