计算机工程与应用 ›› 2018, Vol. 54 ›› Issue (12): 35-40.DOI: 10.3778/j.issn.1002-8331.1704-0357

• 理论与研发 • 上一篇    下一篇

球面l1-正则化逼近模型及其应用研究

陈斯泳,安聪沛   

  1. 暨南大学 信息科学技术学院 数学系,广州 510632
  • 出版日期:2018-06-15 发布日期:2018-07-03

l1-regularization approximation model and its application

CHEN Siyong, AN Congpei   

  1. Department of Mathematics, School of Information Science Technology, Jinan University, Guangzhou 510632, China
  • Online:2018-06-15 Published:2018-07-03

摘要: 基于不同的正则化算子的选取,建立了一类球面上带[l1]-正则项最小二乘逼近模型。通过选取好条件的球面[t]-设计点作为采样点,展示了求解此逼近问题的算法。最后,通过数值例子展现了满意的逼近效果—精确数据和噪声污染的情形。

关键词: 球面多项式逼近, 球面t-设计, l1-正则化

Abstract: Based on variant regularization operators, this paper sets up a class of spherical [l1]-regularization least squares approximation model. The algorithm is illustrated to solve this approximation problem by using well conditioned spherical t-design as sampling point sets. Finally, numerical experiments demonstrate the theoretical results can provide satisfactory approximation on the sphere, with or without the presence of data errors.

Key words: spherical polynomial approximation, spherical [t]-design, [l1]-regularization