Online Feature Selection: A Limited-Memory Substitution
Algorithm and Its Asynchronous Parallel Variation
在线特征选择:有限存储器替换算法及其异步并行变化
ABSTRACT摘要
基于优化算法的在线特征选择有两个缺点:
-
the penalty term for L1 norm term is hard to choose;usually cross validation
L1正则化的罚函数很难确定,一般用交叉验证来确定 -
the memory usage is hard to control or predict
内存很难管理
提出:
a limited-memory and model parameter free online feature selection algorithm, namely online substitution (OS) algorithm. an asynchronous parallel extension for OS (Asy-OS)
有限内存和模型参数游离在线特征选择算法,即在线替换(OS)算法,并且加入异步平行扩展。
关键词:Feature Selection; Online Learning; Asynchronous Parallel
Optimization
1.INTRODUCTION介绍
*two types of online feature selection algorithms:*一般有两种特选算法: