用几何聚合提升个性化联邦学习的准确性和不确定性估计
Personalized Bayesian Federated Learning with Wasserstein Barycenter Aggregation
- 客户端用粒子变分推断实现非参数后验建模
- 服务器采用粒子式沃斯泰因均值聚合，更符合数据几何分布
- 在准确率、校准度和收敛速度上均优于现有方法
个性化贝叶斯联邦学习（PBFL）通过结合个性化与贝叶斯推断，应对非独立同分布的客户端数据并量化不确定性。然而，现有方法存在两个局限：客户端后验推断依赖受限的参数假设，服务器端聚合采用简单的参数平均。为此，我们提出FedWBA，一种新型PBFL方法，同时改进局部推断与全局聚合。客户端采用基于粒子的变分推断，实现非参数后验表示；服务器端引入基于粒子的沃斯泰因均值聚合，提供更几何合理的聚合方式。理论上，我们给出了FedWBA的局部与全局收敛保证：局部上，证明了每轮变分推断中KL散度下降的下界；全局上，表明沃斯泰因均值随客户端数据量增大趋近真实参数。实验显示，FedWBA在预测准确率、不确定性校准和收敛速度上均优于基线方法，消融实验验证其鲁棒性。
原文摘要 · Abstract (English)
Personalized Bayesian federated learning (PBFL) handles non-i.i.d. client data and quantifies uncertainty by combining personalization with Bayesian inference. However, existing PBFL methods face two limitations: restrictive parametric assumptions in client posterior inference and naive parameter averaging for server aggregation. To overcome these issues, we propose FedWBA, a novel PBFL method that enhances both local inference and global aggregation. At the client level, we use particle-based variational inference for nonparametric posterior representation. At the server level, we introduce particle-based Wasserstein barycenter aggregation, offering a more geometrically meaningful approach. Theoretically, we provide local and global convergence guarantees for FedWBA. Locally, we prove a KL divergence decrease lower bound per iteration for variational inference convergence. Globally, we show that the Wasserstein barycenter converges to the true parameter as the client data size increases. Empirically, experiments show that FedWBA outperforms baselines in prediction accuracy, uncertainty calibration, and convergence rate, with ablation studies confirming its robustness.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。