{"title":"基于自适应校准策略的轻量级个性化联邦学习","authors":"Dongshang Deng;Xuangou Wu;Tao Zhang;Chaocan Xiang;Wei Zhao;Minrui Xu;Jiawen Kang;Zhu Han;Dusit Niyato","doi":"10.1109/TSC.2025.3553707","DOIUrl":null,"url":null,"abstract":"Federated learning (FL) is a promising artificial intelligence framework that enables clients to collectively train models with data privacy. However, in real-world scenarios, to construct practical FL frameworks, several challenges have to be addressed, including statistical heterogeneity, constrained resources, and fairness. Therefore, we first investigate an <italic>aggregation gap</i> caused by statistical heterogeneity during local model initialization, which not only causes additional computational overhead for clients but also leads to the degradation of fairness. To bridge this gap, we propose <italic>pFedCal</i>, a novel <underline>p</u>ersonalized <underline>fed</u>erated learning with lightweight adaptive <underline>cal</u>ibration strategy that performs calibration compensation through the prior knowledge of clients. Specifically, we introduce compensation for each client at the model initialization, with the compensation derived from the global gradient and the latest gradient bias. To enhance the calibration effect, we introduce a smoothing-based calibration strategy, and we design an adaptive calibration strategy. A representative example demonstrates that the proposed calibration and smoothing strategies improve fairness for clients. The theoretical analysis indicates that with an appropriate learning rate, pFedCal converges to a first-order stationary point for non-convex loss functions. Comprehensive experimental results show that pFedCal achieves faster convergence, higher accuracy, and improved fairness than the state-of-the-art methods.","PeriodicalId":13255,"journal":{"name":"IEEE Transactions on Services Computing","volume":"18 3","pages":"1627-1640"},"PeriodicalIF":5.8000,"publicationDate":"2025-03-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"pFedCal: Lightweight Personalized Federated Learning With Adaptive Calibration Strategy\",\"authors\":\"Dongshang Deng;Xuangou Wu;Tao Zhang;Chaocan Xiang;Wei Zhao;Minrui Xu;Jiawen Kang;Zhu Han;Dusit Niyato\",\"doi\":\"10.1109/TSC.2025.3553707\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Federated learning (FL) is a promising artificial intelligence framework that enables clients to collectively train models with data privacy. However, in real-world scenarios, to construct practical FL frameworks, several challenges have to be addressed, including statistical heterogeneity, constrained resources, and fairness. Therefore, we first investigate an <italic>aggregation gap</i> caused by statistical heterogeneity during local model initialization, which not only causes additional computational overhead for clients but also leads to the degradation of fairness. To bridge this gap, we propose <italic>pFedCal</i>, a novel <underline>p</u>ersonalized <underline>fed</u>erated learning with lightweight adaptive <underline>cal</u>ibration strategy that performs calibration compensation through the prior knowledge of clients. Specifically, we introduce compensation for each client at the model initialization, with the compensation derived from the global gradient and the latest gradient bias. To enhance the calibration effect, we introduce a smoothing-based calibration strategy, and we design an adaptive calibration strategy. A representative example demonstrates that the proposed calibration and smoothing strategies improve fairness for clients. The theoretical analysis indicates that with an appropriate learning rate, pFedCal converges to a first-order stationary point for non-convex loss functions. Comprehensive experimental results show that pFedCal achieves faster convergence, higher accuracy, and improved fairness than the state-of-the-art methods.\",\"PeriodicalId\":13255,\"journal\":{\"name\":\"IEEE Transactions on Services Computing\",\"volume\":\"18 3\",\"pages\":\"1627-1640\"},\"PeriodicalIF\":5.8000,\"publicationDate\":\"2025-03-24\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"IEEE Transactions on Services Computing\",\"FirstCategoryId\":\"94\",\"ListUrlMain\":\"https://ieeexplore.ieee.org/document/10937074/\",\"RegionNum\":2,\"RegionCategory\":\"计算机科学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q1\",\"JCRName\":\"COMPUTER SCIENCE, INFORMATION SYSTEMS\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"IEEE Transactions on Services Computing","FirstCategoryId":"94","ListUrlMain":"https://ieeexplore.ieee.org/document/10937074/","RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"COMPUTER SCIENCE, INFORMATION SYSTEMS","Score":null,"Total":0}
pFedCal: Lightweight Personalized Federated Learning With Adaptive Calibration Strategy
Federated learning (FL) is a promising artificial intelligence framework that enables clients to collectively train models with data privacy. However, in real-world scenarios, to construct practical FL frameworks, several challenges have to be addressed, including statistical heterogeneity, constrained resources, and fairness. Therefore, we first investigate an aggregation gap caused by statistical heterogeneity during local model initialization, which not only causes additional computational overhead for clients but also leads to the degradation of fairness. To bridge this gap, we propose pFedCal, a novel personalized federated learning with lightweight adaptive calibration strategy that performs calibration compensation through the prior knowledge of clients. Specifically, we introduce compensation for each client at the model initialization, with the compensation derived from the global gradient and the latest gradient bias. To enhance the calibration effect, we introduce a smoothing-based calibration strategy, and we design an adaptive calibration strategy. A representative example demonstrates that the proposed calibration and smoothing strategies improve fairness for clients. The theoretical analysis indicates that with an appropriate learning rate, pFedCal converges to a first-order stationary point for non-convex loss functions. Comprehensive experimental results show that pFedCal achieves faster convergence, higher accuracy, and improved fairness than the state-of-the-art methods.
期刊介绍:
IEEE Transactions on Services Computing encompasses the computing and software aspects of the science and technology of services innovation research and development. It places emphasis on algorithmic, mathematical, statistical, and computational methods central to services computing. Topics covered include Service Oriented Architecture, Web Services, Business Process Integration, Solution Performance Management, and Services Operations and Management. The transactions address mathematical foundations, security, privacy, agreement, contract, discovery, negotiation, collaboration, and quality of service for web services. It also covers areas like composite web service creation, business and scientific applications, standards, utility models, business process modeling, integration, collaboration, and more in the realm of Services Computing.