Summary of Personalized Federated Learning Via Active Sampling, by Alexander Jung and Yasmin Sarcheshmehpour and Amirhossein Mohammadi
Personalized Federated Learning via Active Sampling
by Alexander Jung, Yasmin SarcheshmehPour, Amirhossein Mohammadi
First submitted to arxiv on: 3 Sep 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 This paper proposes a novel method for identifying similar data generators and pooling their local datasets to train personalized machine learning models. The authors address the challenge of training high-dimensional models, such as deep neural networks, when only small local datasets are available from each data generator, which could represent humans with smart-phones or wearables. The proposed method evaluates the relevance of a data generator by analyzing the effect of a gradient step using its local dataset, allowing for privacy-friendly evaluation without sharing raw data. This approach is extended to non-parametric models through a generalization of the gradient step. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Imagine you have many people taking pictures or tracking their fitness with special devices. Each person has some information on their own device, but it’s not enough to train a good machine learning model. The goal is to find similar people and combine their data to get a bigger dataset that can be used for training. This paper presents a new way to do this without sharing the raw data from each person. It works by looking at how well a model does using each person’s data, which helps decide if they are similar or not. This approach can also be applied to other types of models and is designed to be privacy-friendly. |
Keywords
» Artificial intelligence » Generalization » Machine learning » Tracking
