Summary of Data Distribution Shifts in (industrial) Federated Learning As a Privacy Issue, by David Brunner and Alessio Montuoro
Data Distribution Shifts in (Industrial) Federated Learning as a Privacy Issue
by David Brunner, Alessio Montuoro
First submitted to arxiv on: 20 Sep 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 paper explores industrial federated learning, where a few powerful companies collaborate through a third-party mediator to improve their services. However, this setup introduces covert privacy risks that don’t arise in cross-device settings. Companies are protective of their intellectual property and production processes, making it crucial to detect subtle temporal data distribution shifts that might reveal changes to competitors’ production. The authors aim to develop means to better detect these shifts than customary evaluation metrics, which can impact training convergence. They assume minor shifts translate into the collaborative machine learning model’s internal state, allowing an honest-but-curious attacker to track shared models’ states with relevant metrics from literature. In a benchmark dataset study, they demonstrate an attacker’s capability to detect subtle distributional shifts on other clients before becoming obvious in evaluation. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Industrial federated learning is when companies work together through a middleman to make their services better. But this can create secret privacy problems that don’t happen in cross-device settings. Companies want to keep their secrets safe, so it’s important to find out if someone is changing their production process by looking at subtle changes in the data over time. The goal is to develop ways to detect these shifts better than usual methods do. This could help protect companies’ secrets from being discovered. In a test using real datasets, researchers found that an attacker could figure out when someone’s production process changed even before it became obvious. |
Keywords
» Artificial intelligence » Federated learning » Machine learning
