Summary of Analyzing Inference Privacy Risks Through Gradients in Machine Learning, by Zhuohang Li et al.
Analyzing Inference Privacy Risks Through Gradients in Machine Learning
by Zhuohang Li, Andrew Lowy, Jing Liu, Toshiaki Koike-Akino, Kieran Parsons, Bradley Malin, Ye Wang
First submitted to arxiv on: 29 Aug 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR); Machine Learning (stat.ML)
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 Our paper presents a unified game-based framework to analyze private information leakage from gradients in distributed learning settings. We investigate various attacks, including attribute, property, distributional, and user disclosures, and demonstrate the inefficacy of relying solely on data aggregation for privacy against inference attacks. We evaluate five defense mechanisms, including gradient pruning, signed gradient descent, adversarial perturbations, variational information bottleneck, and differential privacy, under both static and adaptive adversary settings. Our results provide an information-theoretic view for analyzing the effectiveness of these defenses. Additionally, we introduce a method for auditing attribute inference privacy and improving empirical estimation of worst-case privacy through crafting adversarial canary records. Our framework encompasses five datasets across various data modalities, demonstrating the versatility and robustness of our approach. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Imagine sharing information with friends, but someone might use that information to figure out your secrets. This paper looks at how to keep those secrets safe when we share information online. We developed a new way to analyze and protect private information from being leaked. We tested this method on different types of data and found that some ways of protecting the information were more effective than others. We also introduced a new tool for checking if someone’s trying to sneakily gather your personal info. Our research can help make online sharing safer and more private. |
Keywords
» Artificial intelligence » Gradient descent » Inference » Pruning
