.png){kind=icon}
Return
Dynamic interaction-based feature selection algorithm for maximal relevance minimal redundancy
.png){kind=icon}
PRE
AI
DOI：10.1007/s10489-022-03922-5.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
In feature selection, distinguishing the redundancy and dependency relationships between features is a challenging task. In recent years, scholars have constantly put forward some solutions, but most of them fail to effectively distinguish dependent features from redundant features. In addition, the influence of feature-relevant complementary item on candidate feature is also ignored, which further reduces the distinguishing ability. In order to improve the distinguishing ability further, the concept of feature interaction degree is proposed, on which the new discriminant criteria of feature redundancy and dependency are defined. With the discriminant criteria and the feature-relevant complementary item, the dynamic interaction weight is constructed. Then a filter feature selection algorithm DIMRMR is proposed to effectively solve the problem of confusing redundancy and dependency. The experimental results shows that the proposed algorithm can achieve the optimal classification performance on most of the datasets.
Keywords:
Feature dependency
Feature redundancy
Mutual information
Filter method
Journal
IF:
3.5
Papers:
7.6K
Citations:
1.7W
Organization
Cited Papers
ShareSave
ShareSaveShareSaveA feature selection algorithm based on redundancy analysis and interaction weight.png)
APPLIED INTELLIGENCE
APPLIED INTELLIGENCEIF3.5
PREAI
ShareSaveShareSaveShareSaveShareSaveShareSaveShareSave
View more