Summary of Federated Learning with Limited Node Labels, by Bisheng Tang et al.
Federated Learning with Limited Node Labels
by Bisheng Tang, Xiaojun Chen, Shaopu Wang, Yuexin Xuan, Zhendong Zhao
First submitted to arxiv on: 18 Jun 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Distributed, Parallel, and Cluster Computing (cs.DC)
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 In this paper, researchers propose a novel framework for subgraph federated learning (SFL) called FedMpa, which addresses limitations in existing SFL models. Specifically, FedMpa aims to learn cross-subgraph node representations by first training a multilayer perceptron (MLP) model with a small amount of data and then propagating the federated feature to local structures. To further improve node embeddings, the authors introduce the FedMpae method, which reconstructs local graph structures using an innovation view that applies pooling operations to form super-nodes. The proposed framework is evaluated on six graph datasets, demonstrating its effectiveness in node classification tasks. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Subgraph federated learning (SFL) is a new way for computers to learn together when they have different pieces of information. Right now, some SFL models are missing an important part: the connections between different groups of data. This can make it hard for computers to share information and work together. To fix this problem, researchers developed a new method called FedMpa that helps computers learn from each other’s information. They also came up with another way to improve how computers represent nodes (important points in the graph) by reconstructing local graph structures. The new methods were tested on six different datasets and showed great results for classifying nodes. |
Keywords
» Artificial intelligence » Classification » Federated learning
