欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
利用稀疏化学习得到的概率图模型结构简单却保留了原始概率图模型中重要的结构信息,且能同时实现结构和参数学习,因此近几年来概率图模型的稀疏化学习一直是研究的热点,其中概率图模型的第一种稀疏化学习方法是图套索.文中总结了概率图模型的稀疏化学习方法,包括概率图模型的L1范数罚稀疏化学习、概率图模型的无偏稀疏化学习、概率图模型的结构稀疏化学习和概率图模型的多任务稀疏化学习.最后,文中还指出了概率图模型的稀疏化学习未来有意义的研究方向.
英文摘要：
A probabilistic graphical model obtained by sparse learning retains the important information of the original probabilistic graphical model＇s structure and the structure of the graphical model after sparse learning is very simple.In addition,the sparse learning can simultaneously achieve the learning of the structure and parameters of the graphical model,so the sparse learning for the probabilistic graphical models has been a research hotspot in recent years and the graphical lasso is the first method for the sparse learning of probabilistic graphical models.This paper summarizes various methods of the sparse learning of probabilistic graphical models,including sparse learning using L1 norm penalty,unbiased sparse learning,sparse structure learning and multi-task sparse learning for probabilistic graphical models.Finally,the paper also proposes the meaningful future research directions for sparse learning of probabilistic graphical models.
同期刊论文项目
同项目期刊论文
A control structure selection method based on multivariable generalized predictive control for unsta
Interaction Analysis and Decomposition Principle for Control Structure Design of Large-scale Systems
期刊信息
位置：