摘要
精度矩阵在高斯图模型中具有广泛的应用,许多正则化方法用于精度矩阵的估计问题,然而却很少有对其解轨迹的关注。本文主要利用条件数正则化协方差矩阵估计的方法来探索精度矩阵估计的解轨迹,数值模拟结果表明该轨迹算法可以便捷地描绘出精度矩阵估计的解轨迹。
Precision matrices are widely used in Gaussian graphical models,and many regularization methods are used to estimate the precision matrix,but there is little attention to solution path of precision matrices estimators.In this paper,we mainly explore the solution path of regularization precision matrix estimators based on a method of condition-number-regularized covariance estimators.Numerical simulation results show that our algorithm can easily paint the solution path of regularization precision matrix estimators.
出处
《安庆师范大学学报(自然科学版)》
2017年第1期49-52,共4页
Journal of Anqing Normal University(Natural Science Edition)
基金
全国统计科学研究项目(2015LZ54)
国家自然科学基金项目(71601003)
关键词
条件数正则化
精度矩阵
解轨迹
轨迹算法
condition-number-regularized
precision matrix
the solution path
path algorithm