Recurrent Transformation of Prior Knowledge Based Model for Human Motion Recognition

Algorithm 1

Divergence-based Feature Selection Algorithm.

Input:  the set of features to be selected

, the variable initialized as 0, and

initialized as .

Output: the set of final selected features

(1)   when   

(2)

(3)   when   

(4)

(5)for   to   do

(6) searching forward in candidate features set

(7)

(8)

(9)

(10)if    then

(11)

(12)break

(13)end if

(14)

(15)if    then

(16)

(17)break

(18)end if

searching backward in selected features set

(19)while TRUE do

(20)

(21)

(22)if    then

(23) no more redundant features, update

(24)

(25)break

(26)end if

(27) roll back selected feature set

(28)

(29)

(30)if    then

(31)

(32)break

(33)end if

(34)if    then

(35)break

(36)end if

(37)end while

(38)end for