Summary of Efficiently Achieving Secure Model Training and Secure Aggregation to Ensure Bidirectional Privacy-preservation in Federated Learning, by Xue Yang et al.
Efficiently Achieving Secure Model Training and Secure Aggregation to Ensure Bidirectional Privacy-Preservation in Federated Learning
by Xue Yang, Depan Peng, Yan Feng, Xiaohu Tang, Weijun Fang, Jun Shao
First submitted to arxiv on: 16 Dec 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 bidirectional privacy-preservation federated learning scheme achieves efficient and high-accuracy model training and aggregation while ensuring local differential privacy (LDP) and central differential privacy (CDP). The scheme consists of two components: MP_Server, a model perturbation method that prevents clients from accessing the model, and DDP_Client, a distributed differential privacy mechanism that ensures LDP of local gradients. Experimental results demonstrate that the scheme outperforms state-of-the-art baselines in terms of computational cost, model accuracy, and defense ability against privacy attacks. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Federated learning is a way for many devices to work together on a task without sharing their data with each other. But this can be a problem if some of those devices don’t want to share their information. This paper proposes a new way to do federated learning that keeps the information private. It’s called bidirectional privacy-preservation, and it has two parts. The first part makes sure that the model being trained doesn’t reveal any secrets about the data it was trained on. The second part makes sure that the local gradients (which are like tiny pieces of information) don’t reveal anything either. This paper shows that this new method is much faster and more accurate than other methods, and it’s better at keeping things private. |
Keywords
» Artificial intelligence » Federated learning
