首页>
外国专利>
L0- Feature selection method using autoregressive model and L0-group lasso and computing system performing the same
L0- Feature selection method using autoregressive model and L0-group lasso and computing system performing the same
展开▼
机译：L0-使用自回归模型和L0-组套索的特征选择方法以及执行该方法的计算系统
展开▼
页面导航
摘要
著录项
相似文献
摘要
Disclosed are a method which uses an autoregressive model and a modified group lasso using an L0-penalty to select a variable more sparsely than an existing algorithm and a variable selection system performing the same. According to an aspect of the present invention, the variable selection method comprises: a step where a variable selection system acquires time-series data of m variables; a step where the variable selection system generates m time-series data groups in accordance with an N^th-order autoregressive model based on the time-series data of m variables; a step where the variable selection system applies a modified group lasso using an L0-penalty to the m time-series data groups to select at least a portion of the m time-series data groups; and a step where the variable selection system determines variables corresponding to the selected at least a portion of the m time-series data groups as main variables.
展开▼
