Summary of Universally Harmonizing Differential Privacy Mechanisms For Federated Learning: Boosting Accuracy and Convergence, by Shuya Feng et al.
Universally Harmonizing Differential Privacy Mechanisms for Federated Learning: Boosting Accuracy and Convergence
by Shuya Feng, Meisam Mohammady, Hanbin Hong, Shenao Yan, Ashish Kundu, Binghui Wang, Yuan Hong
First submitted to arxiv on: 20 Jul 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: 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 The proposed Differentially Private Federated Learning (DP-FL) framework, namely UDP-FL, harmonizes any randomization mechanism with the Gaussian Moments Accountant (DP-SGD) to boost accuracy and convergence. This is achieved by mitigating reliance on Gaussian noise using Rényi Differential Privacy. The framework demonstrates enhanced model performance and superior resilience against inference attacks. Additionally, a novel method for analyzing convergence in DP-FL is proposed based on mode connectivity analysis. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Differentially private federated learning is a way to train models with other people while keeping their data private. This paper proposes a new approach called UDP-FL that makes this process better and more accurate. It does this by using a special kind of noise to protect the privacy of each person’s data, and it works well even when there are different kinds of attacks trying to get at that data. |
Keywords
» Artificial intelligence » Federated learning » Inference
