针对联邦学习中数据异构问题，提出高效通信算法并证明其最优性。
Accelerated Methods with Complexity Separation Under Data Similarity for Federated Learning Problems
- 基于数据相似性建模异构，设计低通信开销优化方法。
- 凸场景下构造出理论最优算法，通信复杂度达下界。
- 适用于数据分布不均的分布式训练场景，如医疗、金融建模。
数据分布异构是现代联邦学习中的主要挑战。本文将其形式化为一个在数据相似性约束下的计算复杂复合优化问题。通过引入不同的假设条件，提出多种通信高效的求解方法。在凸情形下，设计出达到理论最优复杂度的算法。通过在多种任务上的实验验证了所提理论的有效性。
原文摘要 · Abstract (English)
Heterogeneity within data distribution poses a challenge in many modern federated learning tasks. We formalize it as an optimization problem involving a computationally heavy composite under data similarity. By employing different sets of assumptions, we present several approaches to develop communication-efficient methods. An optimal algorithm is proposed for the convex case. The constructed theory is validated through a series of experiments across various problems.
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。