Summary of Vflgan: Vertical Federated Learning-based Generative Adversarial Network For Vertically Partitioned Data Publication, by Xun Yuan and Yang Yang and Prosanta Gope and Aryan Pasikhani and Biplab Sikdar
VFLGAN: Vertical Federated Learning-based Generative Adversarial Network for Vertically Partitioned Data Publication
by Xun Yuan, Yang Yang, Prosanta Gope, Aryan Pasikhani, Biplab Sikdar
First submitted to arxiv on: 15 Apr 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); 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 This paper addresses the challenge of releasing synthetic datasets while preserving correlation among attributes in vertically partitioned data publication. Building upon previous work, VertiGAN, it proposes VFLGAN, a Vertical Federated Learning-based Generative Adversarial Network. Unlike VertiGAN, which is less effective in preserving attribute correlations, VFLGAN significantly improves the quality of synthetic data. The authors demonstrate this improvement using the MNIST dataset, showing that VFLGAN-generated synthetic data has a Fréchet Distance 3.2 times better than VertiGAN’s. Additionally, they develop a Gaussian mechanism for providing differential privacy guarantees and propose a practical auditing scheme to estimate privacy leakage through the synthetic dataset. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Imagine trying to create fake data that looks like real data, but you can’t actually see the real data because it’s private. That’s a big problem in artificial intelligence (AI). This paper proposes a new way to solve this problem by creating “fake” data that is much better than previous methods. They test their method using a famous dataset and show that it produces much higher-quality fake data. They also develop a way to ensure the fake data is private and can’t be used to figure out personal information about individuals. |
Keywords
» Artificial intelligence » Federated learning » Generative adversarial network » Synthetic data
