Summary of Tackling Feature-classifier Mismatch in Federated Learning Via Prompt-driven Feature Transformation, by Xinghao Wu et al.
Tackling Feature-Classifier Mismatch in Federated Learning via Prompt-Driven Feature Transformation
by Xinghao Wu, Jianwei Niu, Xuefeng Liu, Mingjia Shi, Guogang Zhu, Shaojie Tang
First submitted to arxiv on: 23 Jul 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 Personalized Federated Learning (PFL) framework, named FedPFT, addresses the limitations of traditional PFL approaches and surpasses most existing methods, including FedAvg. By integrating a feature transformation module driven by personalized prompts between the global feature extractor and classifier, FedPFT enables clients to train models that better fit their local data distribution while also aligning the training objectives of clients, reducing the impact of data heterogeneity on model collaboration. The framework’s feature transformation module is highly scalable, allowing for the use of different prompts to tailor local features to various tasks. Additionally, a collaborative contrastive learning task is introduced to further refine feature extractor quality. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Federated learning helps machines learn together without sharing personal data. When data is very different from one place to another, it can be hard for the global model to perform well. Personalized federated learning allows each device to train its own model, but most methods aren’t as good as they could be because they compromise on the quality of the feature extractor. The proposed FedPFT framework fixes this issue by transforming local features to match the global classifier and aligning training objectives. This leads to better performance and more accurate results. |
Keywords
» Artificial intelligence » Federated learning
