基于重复测量数据的Tikhonov型正则化方法  

A Tikhonov Regularization Method Based on Repeated Measurement DataSign Language Recognition Algorithm Based on Improved YOLOv5s

作　　者：侯炳雪 董超峰 Hou Bingxu;Dong Chaofeng(School of Mathematical Sciences,Zhejiang Normal University,Jinhua,Zhejiang 321004;College of Data Science,Jiaxing University,Jiaxing,Zhejiang 314001)

机构地区：[1]浙江师范大学数学科学学院,浙江金华321004 [2]嘉兴大学数据科学学院,浙江嘉兴314001

出　　处：《嘉兴大学学报》2024年第6期5-11,共7页Journal of Jiaxing University

摘　　要：针对基于重复测量数据的非线性不适定问题提出了一种Tikhonov型正则化方法:利用“信号平均”重构带有凸罚项的Tikhonov型正则化解,其中正则化参数可利用先验选择规则或序列统计偏差原则确定,并在变分源条件下分析了该方法在这两种参数选择规则下的收敛性结果.This paper proposes a Tikhonov regularization method for nonlinear ill-posed problems based on repeated measurement data.The method reconstructs the Tikhonov regularization solutions with convex penalty terms using“signal averaging”.The regularization parameters are determined by prior selection rules or sequence statistical discrepancy principles.Finally,convergence results of the method based on both parameter selection rules are analyzed under variational source conditions.

关 键 词：多组重复测量数据 Tikhonov型正则化 变分源条件 收敛性分析 

分 类 号：O241[理学—计算数学]