Summary of Post-fair Federated Learning: Achieving Group and Community Fairness in Federated Learning Via Post-processing, by Yuying Duan et al.
Post-Fair Federated Learning: Achieving Group and Community Fairness in Federated Learning via Post-processing
by Yuying Duan, Yijun Tian, Nitesh Chawla, Michael Lemmon
First submitted to arxiv on: 28 May 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Computers and Society (cs.CY)
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 (FL) is a distributed machine learning framework that enables local communities to collaborate on shared global models while retaining training data locally. Recent research highlights the importance of fairness in FL, focusing on group fairness and community fairness. Group fairness ensures model decisions don’t favor specific groups based on legally protected attributes like race or gender, whereas community fairness demands similar performance levels across all participating communities. This paper proposes a post-processing fair federated learning (FFL) framework called post-FFL, which utilizes linear programming to simultaneously enforce group and community fairness while maximizing global model utility. Post-FFL can be applied to existing FL pipelines with well-understood convergence properties. The study demonstrates post-FFL on real-world datasets, mimicking hospital networks using federated learning for community health care. Theoretical results bound the accuracy loss when enforcing both fairness notions, and experimental results show that post-FFL improves both group and community fairness in FL while outperforming existing in-processing FFL methods in terms of communication efficiency and computation cost. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Imagine a world where hospitals can work together to create better health care without sharing sensitive patient data. Federated Learning is a way for groups like this to learn from each other without sharing their own information. Recently, people have realized that it’s important to make sure the models created through this process are fair and don’t unfairly favor certain groups. This paper proposes a new method called post-FFL that helps ensure fairness while also keeping the original model’s performance. The researchers tested this method on real-world data and showed that it works well, even when considering two different types of fairness: group fairness and community fairness. |
Keywords
» Artificial intelligence » Federated learning » Machine learning
