四川大学学报（自然科学版）2025，Vol.62Issue(2)：443-450,8.DOI:10.19907/j.0490-6756.240229
基于相似度感知和选择性通信协议的个性化联邦学习框架
Personalized federated learning framework based on similarity-sense and selective communication protocols
摘要
Abstract
With the advancement of edge computing and wearable technologies,a notable increase in the de-ployment of deep neural networks(DNNs)on endpoint devices has been observed.A personalized federated learning framework named Similarity-Sense Selective Knowledge Distillation(TSKD)is proposed to address the limitations of traditional federated learning frameworks in terms of communication efficiency and model customization.The framework utilizes a small-scale preloaded reference dataset,enabling local user devices to generate communication credentials and assess their similarity with other devices in a heterogeneous net-work.Based on this similarity,devices selectively share knowledge with the most similar devices,thereby en-hancing the performance of their local models.Experiments conducted on three real-world datasets demon-strate that TSKD outperforms traditional centralized and decentralized learning methods across various evalua-tion metrics.Furthermore,TSKD efficiently facilitates knowledge sharing in resource-constrained environ-ments,improving both model accuracy and personalization.关键词
联邦学习/个性化分析/知识蒸馏/数据异质/异构问题Key words
Federated learning/Personalized predictive analytics/Knowledge distillation/Data heterogene-ity/Heterogeneity issues分类
信息技术与安全科学引用本文复制引用
李严,庄孟谕..基于相似度感知和选择性通信协议的个性化联邦学习框架[J].四川大学学报（自然科学版）,2025,62(2):443-450,8.基金项目
国家自然科学基金(62172056) （62172056）
中国人工智能学会青年人才托举工程(2022QNRC001) （2022QNRC001）
