Research Article
Using a Selective Ensemble Support Vector Machine to Fuse Multimodal Features for Human Action Recognition
| Input: The basic classifier set , the diversity threshold , the verification set , the number of base classifiers set and selective ensemble scale N | Process: | (1) While ( { | (2) | (3) | (4) | (5) | (6) The error rates of , , , and on verification set were calculated, with the min error rates saved | , , | , | (7) | (8) if | (9) else if | (10) else if | (11) Else | (12) end} | (13); | Output: The selected base classifier set |
|
|