Oscillation of Metropolis-Hastings and simulated annealing algorithms around LASSO estimator

被引：3
|
作者：
Dermoune, Azzouz [1 ]
Ounaissi, Daoud [1 ]
Rahmania, Nadji [1 ]
机构：
[1] USTL UMR CNRS 8524, Lab Paul Painleve, UFR Math, Bat M2, F-59655 Villeneuve Dascq, France
来源：
MATHEMATICS AND COMPUTERS IN SIMULATION | 2017年 / 135卷
关键词：
Penalized least squares; LASSO; Metropolis Hasting algorithm; Simulated annealing algorithm; Gibbs measures; BAYESIAN COMPUTATION; CONVERGENCE;
D O I：
10.1016/j.matcom.2015.09.003
中图分类号：
TP39 [计算机的应用];
学科分类号：
081203 ; 0835 ;
摘要：
In this work we study, as the temperature goes to zero, the oscillation of a family of Gibbs measures around LASSO estimator. We derive new criteria for estimating LASSO, choosing the proposal distribution and the temperature in Metropolis Hastings algorithm. Finally we apply these results to analyse the convergence of Metropolis Hastings and simulated annealing algorithms. (C) 2015 International Association for Mathematics and Computers in Simulation (IMACS). Published by Elsevier B.V. All rights reserved.
引用
收藏
页码：39 / 50
页数：12
相关论文
共 38 条
1234