Summary of Lid-fl: Towards List-decodable Federated Learning, by Hong Liu et al.
LiD-FL: Towards List-Decodable Federated Learning
by Hong Liu, Liren Shan, Han Bao, Ronghui You, Yuhao Yi, Jiancheng Lv
First submitted to arxiv on: 9 Aug 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 This research proposes an innovative framework for list-decodable federated learning, enabling a central server to maintain a list of models with at least one guaranteed to perform well. The framework relaxes the requirement for honest worker participation, making Byzantine federated learning more practical in scenarios where over half of participants might be adversaries. Under specific assumptions on the loss function, the authors prove a convergence theorem for their method. Experimental results, featuring image classification tasks with both convex and non-convex losses, demonstrate the algorithm’s resilience against malicious attacks. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This study develops an algorithm to help groups work together on a task while keeping some members’ contributions trustworthy. Imagine you’re part of a team working on a project, but some members might not be telling the truth about their work. This paper shows how to design a system where at least one member’s contribution is reliable, even if most others are trying to deceive you. The researchers tested their idea with image classification tasks and found that it can handle attacks from malicious team members. |
Keywords
» Artificial intelligence » Federated learning » Image classification » Loss function
