Summary of Fedcar: Cross-client Adaptive Re-weighting For Generative Models in Federated Learning, by Minjun Kim et al.
FedCAR: Cross-client Adaptive Re-weighting for Generative Models in Federated Learning
by Minjun Kim, Minjee Kim, Jinhoon Jeong
First submitted to arxiv on: 16 Dec 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
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 Generative models trained on multi-institutional datasets can provide an enriched understanding through diverse data distributions. However, training the models on medical images is often challenging due to hospitals’ reluctance to share data for privacy reasons. Federated learning (FL) has emerged as a privacy-preserving solution for training distributed datasets across data centers by aggregating model weights from multiple clients instead of sharing raw data. To improve the performance of generative models within FL, we propose a novel algorithm that adaptively re-weights the contribution of each client, resulting in well-trained shared parameters. Our approach also enhances efficiency by measuring distribution distance between fake images generated by clients rather than directly comparing Fréchet Inception Distance per client. Experimental results on three public chest X-ray datasets show superior performance in medical image generation, outperforming both centralized learning and conventional FL algorithms. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Generative models can learn from a variety of data sources, but sharing medical images for training can be challenging due to privacy concerns. Federated learning is a way to train models without sharing the actual data. Researchers have been exploring how to use federated learning with generative models, but there isn’t an effective algorithm yet. Our new approach reweights the contributions of different groups providing data and measures the distance between fake images they generate to improve training efficiency. We tested our method on three public chest X-ray datasets and found it outperforms other methods. |
Keywords
» Artificial intelligence » Federated learning » Image generation
