Summary of Fedspu: Personalized Federated Learning For Resource-constrained Devices with Stochastic Parameter Update, by Ziru Niu et al.
FedSPU: Personalized Federated Learning for Resource-constrained Devices with Stochastic Parameter Update
by Ziru Niu, Hai Dong, A. K. Qin
First submitted to arxiv on: 18 Mar 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 The proposed Federated Learning with Stochastic Parameter Update (FedSPU) outperforms Federated Dropout by 7.57% on average in terms of accuracy, addressing performance degradation caused by biased sub-models absorbing highly divergent parameters from other clients. FedSPU maintains the full model architecture on each device, randomly freezing a certain percentage of neurons during training while updating the remaining neurons. This approach ensures the model’s robustness against biased parameters, reducing computation and communication overheads in IoT applications. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Federated learning helps devices share information without sharing their data. But when different devices have limited resources, it can be slow and inaccurate. The authors propose a new way to do this called Federated Learning with Stochastic Parameter Update (FedSPU). It’s faster and more accurate than the old method by 7.57%. They also came up with an early stopping scheme that saves time while keeping accuracy high. |
Keywords
* Artificial intelligence * Dropout * Early stopping * Federated learning
