Summary of Differentially Private Decentralized Learning with Random Walks, by Edwige Cyffers et al.
Differentially Private Decentralized Learning with Random Walks
by Edwige Cyffers, Aurélien Bellet, Jalaj Upadhyay
First submitted to arxiv on: 12 Feb 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 research characterizes the privacy guarantees of decentralized learning using random walk algorithms, which update a model by traversing a communication graph. By applying Pairwise Network Differential Privacy to this setting, the study derives closed-form expressions for privacy loss between nodes, capturing the impact of the graph topology. Results show that random walk algorithms tend to provide better privacy guarantees than gossip algorithms for nodes in close proximity. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper explores how decentralized learning with random walk algorithms can protect personal data while sharing model updates. By using a special type of differential privacy called Pairwise Network Differential Privacy, researchers found ways to calculate the privacy level between different nodes on a communication graph. They also compared this approach to another method called gossip algorithm and discovered that random walk is better at keeping data private for nearby nodes. |
