Summary of Kdk: a Defense Mechanism Against Label Inference Attacks in Vertical Federated Learning, by Marco Arazzi et al.
KDk: A Defense Mechanism Against Label Inference Attacks in Vertical Federated Learning
by Marco Arazzi, Serena Nicolazzo, Antonino Nocera
First submitted to arxiv on: 18 Apr 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 framework, KDk, combines Knowledge Distillation and k-anonymity to defend against label inference attacks in Vertical Federated Learning (VFL) scenarios. By leveraging gradient information and auxiliary labels, attackers can infer private labels from a limited subset of training data points. The authors demonstrate that their approach significantly decreases the performance of these attacks by over 60% while maintaining the overall VFL accuracy. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary In this paper, scientists developed a new way to keep people’s private information safe in Vertical Federated Learning (VFL). They showed that others can figure out private labels if they have some extra information. To fix this problem, they created a special framework called KDk, which combines two existing methods. This helps protect VFL by making it harder for attackers to find private labels. The results are impressive, showing a big decrease in the effectiveness of these attacks. |
Keywords
» Artificial intelligence » Federated learning » Inference » Knowledge distillation
