百度文库 其他 adaptive personalized federated learningadaptive personalized federated learning adaptive personalized federated learning翻译为:自适应个性化联邦学习。©2022 Baidu |由 百度智能云 提供计算服务 | 使用百度前必读 | 文库协议 | 网站地图 | 百度营销 ...
作者认为最大化 global model 的性能会影响其在 local data 上的泛化能力,本文提出 adaptive personalized federated learning (APFL),每个 local device 训练 local model 的同时为 global model 做出贡献。并且从理论上分析了个性化模型对 local data 的泛化能力,依赖于混合参数、局部分布和全局分布之间的差异以及局部...
本地微调(Local fine tuning):本地微调即每个客户端接收到一个全局模型,并使用自己的局部数据和几个梯度下降步骤对其进行调优,这种方法主要结合了元学习。 多任务学习(multi_task learning):对个性化问题的另一种观点是视为多任务学习问题。这种设置下对每个客户端的优化可以看做是一个新的任务。 情景化(Contextualiz...
Investigation of the degree of personalization in federated learning algorithms has shown that only maximizing the performance of the global model will confine the capacity of the local models to personalize. In this paper, we advocate an adaptive personalized federated learning (APFL) algorithm, where...
Personalized federated learning (PFL) is an improved framework that can facilitate the handling of data heterogeneity by learning personalized models. As personalization performance directly depends on the global model, it is desired to acquire a global model with a decent generalization capability under...
Adaptive Personalized Over-the-Air Federated Learning with Reflecting Intelligent Surfaces 来自 arXiv.org 喜欢 0 阅读量: 1 作者:Mao, Jiayu,Yener, Aylin 摘要: Over-the-air federated learning (OTA-FL) unifies communication and model aggregation by leveraging the inherent superposition property of the...
Adaptive Test-Time Personalization for Federated Learning. NeurIPS 2023. [Arxiv] [Poster] [Slides] Introduction We consider a novel setting named Test-Time Personalized Federated Learning, addressing the challenge of personalizing a global model to each unparticipating client during test-time, without ...
This is the implementation of our paperFedALA: Adaptive Local Aggregation for Personalized Federated Learning(accepted by AAAI 2023). An extended version (derivation of Equation (6), hyperparameter settings, etc.) can be found athttps://arxiv.org/pdf/2212.01197v4.pdf. ...
However, the parameter- efficient method of freezing the backbone network is con- sistent with the mechanism of personalized FL which is also an effective way to improve FL performance. More im- portantly, the number of training parameters and commu- nication cost of our met...
Song, J., Xu, J., Zhou, R., Chen, L., Li, J., Liu, C.: CBML: A cluster-based meta-learning model for session-based recommendation. In: CIKM, pp 1713–1722 (2021) Google Scholar Yu, R., Gong, Y., He, X., Zhu, Y., Liu, Q., Ou, W., An, B.: Personalized adaptiv...