Summary of Enhancing Privacy in Federated Learning Through Local Training, by Nicola Bastianello et al.
Enhancing Privacy in Federated Learning through Local Training
by Nicola Bastianello, Changxin Liu, Karl H. Johansson
First submitted to arxiv on: 26 Mar 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Optimization and Control (math.OC)
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 proposes Fed-PLT, a federated learning algorithm that tackles two challenges in distributed machine learning: expensive communications and privacy preservation. To address these issues, Fed-PLT allows for partial participation and local training, reducing the number of communication rounds while maintaining accuracy. The algorithm also enables agents to choose from various local training solvers like gradient descent and accelerated gradient descent. Furthermore, the paper investigates how local training can enhance privacy by deriving differential privacy bounds and highlighting their dependence on the number of local training epochs. The effectiveness of Fed-PLT is assessed through both theoretical analysis and numerical results in a classification task. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper creates a new way for computers to learn together without sharing too much information. It’s like a team effort, where each computer does some learning on its own before sharing with others. This makes it faster and more private. The new method is called Fed-PLT. It allows each computer to choose how it learns best and still gets good results. The paper also shows that this way of learning can be made even more private by controlling how much information is shared. |
Keywords
* Artificial intelligence * Classification * Federated learning * Gradient descent * Machine learning
