FedCRAC: Improving Federated Classification Performance on Long-Tailed Data via Classifier Representation Adjustment and Calibration

doi:10.1109/TMC.2024.3466208

	FedCRAC: Improving Federated Classification Performance on Long-Tailed Data via Classifier Representation Adjustment and Calibration
	Li, Xujing 1,2; Sun, Sheng 1; Liu, Min 2,3,4; Ren, Ju 5; Jiang, Xuefeng 1,2; He, Tianliu 1,2
	2025
发表期刊	IEEE TRANSACTIONS ON MOBILE COMPUTING
ISSN	1536-1233
卷号	24 期号:1 页码:482-499
摘要	Federated learning has been a popular distributed training paradigm that enables to train a shared model with data privacy protection. However, non-Independent Identically Distribution and long-tailed data distribution characteristics across mobile devices results in evident performance degradation, especially for classification tasks. Although plenty of research studies devote to alleviating classification performance degradation caused by highly-skewed data distribution, they still cannot improve the distinguishability of model representation on hard-to-learn tail classes, and face obvious divergence of local classifiers in FL setting. To this end, we propose Federated Classifier Representation Adjustment and Calibration to improve the representation distinguishability of tail classes and achieve inter-client representation alignment with acceptable resource consumption on attaching operations. We first design a Class Similarity-Aware Margin matrix to enlarge class representation discrepancy and improve local classifier discriminability on tail classes during client-side local training process. To mitigate the divergence of local classifiers across clients, we further propose the Self Distillation Classifier Calibration to achieve the aggregated global classifier calibration with the assistance of generated pseudo representation samples via self-distillation manner. We conduct various experiments under wide-range long-tailed and heterogeneous data settings. Experimental results show that FedCRAC outperforms state-of-the-art methods in terms of accuracy and resource consumption.
关键词	Tail Data models Training Computational modeling Servers Accuracy Feature extraction Data heterogeneity federated learning long-tailed data representation alignment
DOI	10.1109/TMC.2024.3466208
收录类别	SCI
语种	英语
资助项目	National Key Research and Development Program of China[2021YFB2900102] ; National Natural Science Foundation of China[62072436]
WOS研究方向	Computer Science ; Telecommunications
WOS类目	Computer Science, Information Systems ; Telecommunications
WOS记录号	WOS:001370229700018
出版者	IEEE COMPUTER SOC
引用统计
文献类型	期刊论文
条目标识符	http://119.78.100.204/handle/2XEOYT63/41119
专题	中国科学院计算技术研究所期刊论文_英文
通讯作者	Liu, Min
作者单位	1.Chinese Acad Sci, Inst Comp Technol, Beijing 100045, Peoples R China 2.Univ Chinese Acad Sci, Beijing 101408, Peoples R China 3.Chinese Acad Sci, Inst Comp Technol, SKLP, Beijing 100045, Peoples R China 4.Zhongguancun Lab, Beijing 100190, Peoples R China 5.Tsinghua Univ, Dept Comp Sci & Technol, BNRist, Beijing 100190, Peoples R China
推荐引用方式 GB/T 7714	Li, Xujing,Sun, Sheng,Liu, Min,et al. FedCRAC: Improving Federated Classification Performance on Long-Tailed Data via Classifier Representation Adjustment and Calibration[J]. IEEE TRANSACTIONS ON MOBILE COMPUTING,2025,24(1):482-499.
APA	Li, Xujing,Sun, Sheng,Liu, Min,Ren, Ju,Jiang, Xuefeng,&He, Tianliu.(2025).FedCRAC: Improving Federated Classification Performance on Long-Tailed Data via Classifier Representation Adjustment and Calibration.IEEE TRANSACTIONS ON MOBILE COMPUTING,24(1),482-499.
MLA	Li, Xujing,et al."FedCRAC: Improving Federated Classification Performance on Long-Tailed Data via Classifier Representation Adjustment and Calibration".IEEE TRANSACTIONS ON MOBILE COMPUTING 24.1(2025):482-499.