Summary of Fligan: Enhancing Federated Learning with Incomplete Data Using Gan, by Paul Joe Maliakel et al.
FLIGAN: Enhancing Federated Learning with Incomplete Data using GAN
by Paul Joe Maliakel, Shashikant Ilager, Ivona Brandic
First submitted to arxiv on: 25 Mar 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: None
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 Federated Learning (FL) enables AI at the edge by training machine learning models on networked devices without sharing actual data. While existing research focuses on non-IID data and heterogeneity, it often neglects insufficient data for model development due to uneven class label distribution and variable data volumes across nodes. This paper proposes FLIGAN, a novel approach that addresses data incompleteness in FL by leveraging Generative Adversarial Networks (GANs) to generate synthetic data resembling real-world data. Synthetic data enhances dataset robustness and completeness, adhering to FL’s privacy requirements. Techniques like classwise sampling and node grouping improve federated GAN performance, enabling high-quality synthetic datasets and efficient FL training. Experimental results demonstrate up to a 20% increase in model accuracy over traditional FL baselines, especially in scenarios with high class imbalances. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Imagine being able to train AI models on your phone or other devices without sharing any of your personal data. This paper proposes a new way to do just that, called Federated Learning (FL). Currently, when we train AI models, we share our data with others, which can be a problem for privacy. The new approach uses special computer programs called Generative Adversarial Networks (GANs) to create fake data that looks like real data. This fake data is used to improve the quality of the training data and make the AI models more accurate. The results show that this approach works really well, especially when there’s not enough data to train a good model. |
Keywords
* Artificial intelligence * Federated learning * Gan * Machine learning * Synthetic data
