MAP: Model Aggregation and Personalization in Federated Learning with Incomplete Classesarxiv.org/abs/2404.09232
论文题目:《FedSim: Similarity guided model aggregation for Federated Learning》论文发表在期刊《Neurocomputing》上,等级CCF-C ;所属单位:Robert Gordon University, UK;Accepted 30 August 2021。 1.背景 联邦学习分跨孤岛(cross-silo)和跨设备(cross-device)两种应用场景。跨孤岛一般是指拥有数据中心的不同组织...
In this article, accuracy-based federated learning with adaptive model aggregation (A-FLAMA) is introduced. With this algorithm, client models that achieve higher accuracy with the server's test data are assigned heavier aggregation weights. This eliminates the need for extra information from clients...
Section 2 summarizes related works on improving the efficiency of federated learning. In Section 3, we introduce a preliminary study on partial model aggregation, which shows the resulting accuracy when the server only aggregates a part of participants in a round and shows the influence factors to...
FAIR: Quality-Aware Federated Learning with Precise User Incentive and Model Aggregation阅读笔记 阅读目录(Content) 动机 算法 问题描述 学习质量估计 质量感知激励机制 模型聚合 结果 总结与思考 回到顶部(go to top) 动机本文是2021年infocom上的一篇论文。联邦学习面临着两大挑战:1.用户可能并不愿意参与到学习...
Model aggregation and client-side data encryption are used to combat this problem. The Federated learning protocol combines and sums the model output, and the server has access only to the aggregate model and not the individual models. Here, the devices report only the data that is required ...
初识Federated Learning 背景 设备中有很多数据,可以用来训练模型提高用户体验。但是数据通常是敏感或者庞大的。 隐私问题 数据孤岛:每个公司都有数据,淘宝有你的购买记录,银行有你的资金状况,它们不能把数据共享,都是自己训练自己有的数据,是一个个数据孤岛。 联邦学习的概念 联邦机器学习是一个机器学习框架,能有效...
2016年,为解决安卓系统更新的问题。谷歌提出,可以在用户的手机上部署神经网络训练,只需要将训练好的模型参数上传,而不需要上传用户数据,一定程度上保证了个人数据的私密。这就是联邦学习(federated learning)的核心理念。 (2)不同使用场景 各个参与者的业务类型相似,数据特征重叠多,样本重叠少。例如:不同地区的两家银...
MocoSFL: enabling cross-client collaborative self-supervised learning Multimodal Federated Learning via Contrastive Representation Ensemble Panning for Gold in Federated Learning: Targeted Text Extraction under Arbitrarily Large-Scale Aggregation PerFedMask: Personalized Federated Learning with Optimized Masking Vect...
FMore是轻量级框架,兼容性好,计算和通信的开销低。《2021-IEEE-FAIR: Quality-Aware Federated Learning with Precise User Incentive and Model Aggregation》提出了一种质量感知拍卖方法,将获胜者选择问题描述为一个 NP 难的学习质量最大化问题,并基于迈尔森定理设计了一个贪婪算法来执行实时任务分配和报酬分配。