Using a Selective Ensemble Support Vector Machine to Fuse Multimodal Features for Human Action Recognition

Table 1

SESVM.

---

Input:

Training set , verification set , base classification algorithm SVM, number of base classifiers , number of selected base classifiers

Output:

Selected base classifier set

Training process:

(1) Initialize the base classifier set

(2) For

(3) Based on the training set , a new training set is obtained by using Bootstrap random sampling method

(4) The base classifier is trained on the training set by using the base classification algorithm and added to the set

(5) End for

(6) Selecting process:

(7) Each base classifier is tested on verification set and its output is obtained

(8) The selected base classifier set is obtained by using CCCSA

---