Imbalanced Learning Based on Logistic Discrimination

Algorithm 1

ILLD.

Input:

: training data set

: parameter greater than zero

Output:

learned parameters

Process:

(1) randomly initialize ;

(2)  (unit matrix);

(3)  APM // calculate the gradient of object function by (14)

(4)  // update using (15)

(5)  ;

(7)  while   

(8)  Calculate the gradients of as using (1)

(9)  update and using (17);

(10)  update using (16);

(11)    ; // update using (15)

(12)

(13) return