Abstract
Federated Learning (FL) has emerged as a prominent privacy-preserving technique for enabling use cases like confidential clinical machine learning. FL operates by aggregating models trained by remote devices which owns the data. Thus, FL enables the training of powerful global models using crowd-sourced data from a large number of learners, without compromising their privacy. However, the aggregating server is a single point of failure when generating the global model. Moreover, the performance of the model suffers when the data is not independent and identically distributed (non-IID data) on all remote devices. This leads to vastly different models being aggregated, which can reduce the performance by as much as 50% in certain scenarios. In this paper, we seek to address the aforementioned issues while retaining the benefits of FL. We propose MultiConfederated Learning: a decentralized FL framework which is designed to handle non-IID data. Unlike traditional FL, MultiConfederated Learning will maintain multiple models in parallel (instead of a single global model) to help with convergence when the data is non-IID. With the help of transfer learning, learners can converge to fewer models. In order to increase adaptability, learners are allowed to choose which updates to aggregate from their peers.
Abstract (translated)
联邦学习(FL)已成为一种突出隐私保护的技术,用于实现诸如机密临床机器学习等隐私 preserving 用例。FL通过汇总由远程设备训练的模型来操作,这些设备拥有数据。因此,FL允许大规模学习者使用 crowd-sourced 数据训练强大的全局模型,同时不损害他们的隐私。然而,在生成全局模型时,聚合服务器是一个单点故障。此外,当数据不独立且分布不同时(非 IID 数据)在所有远程设备上训练模型时,模型的性能会受到影响。这导致聚合的不同模型,在某些场景下可能导致性能降低 50%。在本文中,我们试图解决上述问题,同时保留 FL 的优点。我们提出了 MultiConfederated Learning:一个设计用于处理非 IID 数据的联邦 FL 框架。与传统 FL 不同,MultiConfederated Learning 将多个模型并行(而不是单全球模型)以帮助数据为非 IID 时达到收敛。通过迁移学习,学习者可以 convergence 到更少的模型。为了增加适应性,学习者被允许从同伴中选择要聚合的更新。
URL
https://arxiv.org/abs/2404.13421