federated learning
PulseAugur coverage of federated learning — every cluster mentioning federated learning across labs, papers, and developer communities, ranked by signal.
- 2026-05-22 research_milestone Publication of a paper detailing an embedding-based federated learning system for iron deficiency prediction. 来源
9 天有情绪数据
-
New framework 'Mechanical Conscience' offers trajectory-level regulation for AI
A new paper introduces "mechanical conscience" (MC), a mathematical framework designed to regulate the behavior of intelligent systems, particularly in distributed collaborative intelligence (DCI) environments. This fra…
-
AutoFLIP framework harnesses client diversity to prune federated models efficiently
Researchers have developed AutoFLIP, a new framework designed to improve the efficiency of Federated Learning (FL) on devices with limited resources. This approach leverages the diversity of client data, rather than tre…
-
分层联邦学习框架重新定义网络化AI设计
本文提出分层联邦学习（HFL）作为一种面向架构感知的网络化AI设计框架，超越了其作为通信节省协议的常见表述。作者认为，HFL应围绕三个轴组织：架构参数、层级优化分解和层级通信实现。他们证明了HFL中的收敛性依赖于架构，并受所选层级、优化角色和通信机制的影响。
-
Federated Learning benchmark introduced for adaptation, trust, and reasoning
A new benchmark framework called ATR-Bench has been proposed to standardize the evaluation of Federated Learning (FL) techniques across adaptation, trust, and reasoning. The paper details conceptual foundations and task…
-
New research explores federated learning vulnerabilities and defenses against backdoor attacks
Researchers have developed new methods to combat sophisticated backdoor attacks in federated learning. One approach, DeTrigger, uses gradient analysis to detect and remove malicious triggers with minimal impact on model…
-
Victor Chang教授因在网络安全和隐私领域的AI领导力而获奖
Victor Chang教授因其在负责任的AI开发方面所做的贡献，被评为年度网络安全专业人士。他的工作强调联邦学习和隐私保护技术，特别是在保护关键基础设施方面。Chang教授牵头了一个英日合作项目，该项目开发了一个联邦恶意软件检测系统，准确率达到96.8%，同时保护了用户隐私。
-
FedPLT 通过部分层训练提供资源高效的联邦学习
研究人员推出了一种新颖的联邦学习方法 FedPLT，该方法旨在实现可扩展、资源高效且能适应异构环境。该方法仅在单个客户端上训练模型的特定层，并根据客户端的计算和通信能力进行定制。FedPLT 旨在实现与完整模型训练相媲美的性能，同时显著减少每个客户端的可训练参数数量，有望克服去中心化机器学习中的通信和计算开销。
-
FedACT optimizes concurrent federated learning across heterogeneous devices
Researchers have developed FedACT, a new resource-aware scheduling approach for federated learning systems. This method aims to improve efficiency and reduce job completion times when multiple machine learning tasks are…
-
New framework uses K-Shapley values for meritocratic fairness in bandits
Researchers have introduced a novel framework for achieving meritocratic fairness in budgeted combinatorial multi-armed bandits with full-bandit feedback. This new approach extends the Shapley value concept to a K-Shapl…
-
Researchers propose AdaBFL for robust federated learning against attacks
Researchers have introduced AdaBFL, a novel multi-layer defensive aggregation method designed to enhance the robustness of federated learning against Byzantine attacks. This approach addresses limitations of existing me…
-
New framework enables asynchronous federated unlearning for medical imaging models
Researchers have introduced Asynchronous Federated Unlearning with Invariance Calibration (AFU-IC), a new framework designed for medical imaging applications. This method addresses limitations in existing Federated Unle…
-
Federated learning framework tackles medical imaging imbalance with synthetic data
Researchers have developed FedSSG, a new Federated Learning framework designed to improve medical image classification. This framework addresses challenges like data privacy, varying imaging device properties, and imbal…
-
新研究推动联邦学习在隐私和异构性方面的进展
研究人员正在开发新的方法来改进联邦学习，这是一种允许模型在不损害隐私的情况下对去中心化数据进行训练的技术。几篇论文介绍了处理数据异构性的新算法，例如用于随机森林的FedForest和用于物联网系统中客户端选择的VARS-FL。其他工作侧重于通过共识嵌入进行隐私保护推理以及用于联邦图神经网络的鲁棒方法。此外，正在探索新的理论框架来限制泛化误差并激励联邦环境中的客户端贡献。
-
New DSFL framework enhances scalable and verifiable financial fraud detection
Researchers have introduced Dynamic Sharded Federated Learning (DSFL), a new framework designed to enhance cross-institutional financial fraud detection while preserving data privacy. DSFL addresses limitations in exist…
-
Federated learning paper introduces new strategy for client disagreements
This paper introduces a new taxonomy and resolution strategy for handling client-level disagreements in Federated Learning (FL). The proposed method creates isolated model update paths to prevent cross-contamination and…
-
联邦学习进展在隐私、效用和公平性之间取得平衡
研究人员正在探索增强联邦学习（FL）隐私的高级技术，FL是一种在去中心化数据上进行模型训练的方法。一项研究比较了在瑞典医疗保健数据上用于心血管疾病风险建模的差分隐私（DP）和同态加密（HE），发现HE与集中式方法相当，但计算开销更高，而DP在某些模型上表现出更大的性能下降。另一种方法FedPF引入了一种差分隐私的公平FL算法，通过将公平性和效用视为竞争目标来平衡它们，在具有竞争力的准确性和低计算占用的情况下显著减少了歧视。第三篇论文将…
-
Federated Learning uses spectral entropy for data-free client contribution estimation
Researchers have developed a novel method for estimating client contributions in Federated Learning without requiring access to client data. This approach utilizes the spectral entropy of final-layer updates to measure …
-
Apple 详解用于 Apple Intelligence 的隐私保护 AI 研究和差分隐私
Apple 正在推进隐私保护机器学习和 AI 的研究，并举办研讨会讨论联邦学习和差分隐私等技术。该公司正在将其即将推出的 Apple Intelligence 功能（如 Genmoji、Image Playground 和写作工具）应用于这些方法，以了解使用趋势，同时不损害用户数据。Apple 还在探索创建模仿真实用户内容的合成数据，以在保持严格隐私标准的同时改进这些功能。