Research Article
A Feature Selection Algorithm Integrating Maximum Classification Information and Minimum Interaction Feature Dependency Information
| (1) | Input: Original feature set ; Class label set ; Threshold | | (2) | Output: Optimal feature subset | | (3) | initialization: ; | | (4) | fordo | | (5) | Calculate the mutual information value of each feature and label ; | | (6) | ifthen | | (7) | remove from and continue; | | (8) | end | | (9) | end | | (10) | ; | | (11) | ; | | (12) | ; | | (13) | whiledo | | (14) | calculate the value of ; | | (15) | ifthen | | (16) | calculate the value of ; | | (17) | calculate the value of ; | | (18) | Update using equation (10); | | (19) | find the candidate feature with the largest ; | | (20) | end | | (21) | ; | | (22) | | | (23) | ; | | (24) | end |
|
