Summary of Eab-fl: Exacerbating Algorithmic Bias Through Model Poisoning Attacks in Federated Learning, by Syed Irfan Ali Meerza et al.
EAB-FL: Exacerbating Algorithmic Bias through Model Poisoning Attacks in Federated Learning
by Syed Irfan Ali Meerza, Jian Liu
First submitted to arxiv on: 2 Oct 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR)
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 focuses on Federated Learning (FL), a technique that enables multiple parties to train a shared model collaboratively without sharing their private data. Despite its advantages in privacy, FL models can suffer from biases against certain demographic groups due to heterogeneous data and party selection. To address this issue, researchers have proposed various strategies for characterizing the group fairness of FL algorithms. However, the effectiveness of these strategies has not been fully explored in the face of deliberate adversarial attacks. The paper proposes a new type of model poisoning attack, EAB-FL, which focuses on exacerbating group unfairness while maintaining good model utility. Extensive experiments on three datasets demonstrate the effectiveness and efficiency of this attack even with state-of-the-art fairness optimization algorithms and secure aggregation rules employed. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This research is about making sure that machines can learn together without sharing their private information. Sometimes, when they do this, they can end up being unfair to certain groups of people. To fix this problem, scientists have tried different ways to make sure the machine learning is fair. However, nobody has looked at what happens when someone intentionally tries to make it unfair. This paper proposes a new way to make the machines learn unfairly while still doing their job well. The researchers tested this idea on three groups of data and found that it works even with special tools designed to keep things fair. |
Keywords
» Artificial intelligence » Federated learning » Machine learning » Optimization
