Distance-based feature selection from probabilistic data

摘要: Feature selection is a powerful tool of dimension reduction from datasets. In the last decade, more and researchers have paid attentions on feature selection. Further, some begin to focus probabilistic However, in existing method data, distance hidden data neglected. this paper, we design new measure select informative databases, which both randomness are considered. And then, propose algorithm based develop two accelerative algorithms boost computation. Furthermore, introduce parameter into reduce sensitivity noise. Finally, experimental results verify effectiveness our algorithms.