Summary of Fedsi: Federated Subnetwork Inference For Efficient Uncertainty Quantification, by Hui Chen et al.
FedSI: Federated Subnetwork Inference for Efficient Uncertainty Quantification
by Hui Chen, Hengyu Liu, Zhangkai Wu, Xuhui Fan, Longbing Cao
First submitted to arxiv on: 24 Apr 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI)
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 federated learning (FL) framework, called FedSI, which addresses the issue of efficient systematic uncertainty quantification in personalized FL. FedSI is based on Bayesian deep neural networks (DNNs) and leverages their ability to incorporate uncertainties effectively. The framework implements a client-specific subnetwork inference mechanism, selecting network parameters with large variance to be inferred through posterior distributions, while fixing the rest as deterministic ones. This approach enables fast and scalable inference while preserving systematic uncertainties. FedSI outperforms existing FL baselines in heterogeneous scenarios on three benchmark datasets. |
Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper develops a new way for computers to learn together using different types of data. They call it FedSI, which is short for “federated subnetwork inference”. It’s like having many small versions of the same computer program, each one special and unique. The new method helps figure out what parts of the program need more attention by looking at how uncertain they are. This allows the computers to work together better when their data is different. Tests showed that this approach works well on three different types of datasets. |
Keywords
» Artificial intelligence » Attention » Federated learning » Inference