Robust and Personalized Federated Learning with Spurious Features: an Adversarial Approach
Robust and Personalized Federated Learning with Spurious Features: an Adversarial Approach
复制标题
DOI：
--
复制
发表时间：
--
期刊：
影响因子：
--
通讯作者：
Xiaoyang Wang;Han Zhao;K. Nahrstedt;Sanmi Koyejo
中科院分区：
文献类型：
--
作者：
Xiaoyang Wang;Han Zhao;K. Nahrstedt;Sanmi Koyejo
The most common approach for personalized federated learning is fine-tuning the global machine learning model to each client. While this addresses some issues of statistical diversity, we find that such personalization methods are vulnerable to spurious features , leading to bias and sacrificing generalization. Nevertheless, debiasing the personalized models is difficult. To this end, we propose a strategy to mitigate the effect of spurious features based on an observation that the global model in the federated learning step has a low bias degree due to statistical diversity. Then, we estimate and mitigate the bias degree difference between the personalized and global models using adversarial transferability in the personalization step. Empirical results on MNIST, CelebA, and Coil20 datasets show that our method improves the accuracy of the personalized model on the bias-conflicting data samples by up to 14.3%, compared to existing personalization approaches, while preserving the benefit of enhanced average accuracy from fine-tuning.
