Summary of Debiasing Federated Learning with Correlated Client Participation, by Zhenyu Sun et al.
Debiasing Federated Learning with Correlated Client Participation
by Zhenyu Sun, Ziyang Zhang, Zheng Xu, Gauri Joshi, Pranay Sharma, Ermin Wei
First submitted to arxiv on: 2 Oct 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Distributed, Parallel, and Cluster Computing (cs.DC)
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 introduces a theoretical framework to analyze optimization convergence in federated learning (FL) when clients have non-uniform and correlated participation across rounds. Specifically, it models client participation as a Markov chain and studies how increasing the minimum time between client participations affects bias induction in cross-device FL systems. The authors theoretically prove and empirically observe that this increase reduces bias, and develop an effective debiasing algorithm for Federated Averaging (FedAvg) that provably converges to the unbiased optimal solution. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper is about making sure computers learn together better when they don’t all participate at the same time. Usually, only a few of the many devices involved in learning do it every time, and this can affect how well they learn. The researchers created a way to understand what happens when some devices have to wait before participating again, which is common in real-world scenarios. They found that making these devices wait longer reduces the mistakes made by those who don’t participate as often. They also came up with an algorithm to fix this problem and make sure all devices learn equally well. |
Keywords
» Artificial intelligence » Federated learning » Optimization
