Cognitive Driven Multilayer Self-Paced Learning with Misclassified Samples

Algorithm 1

The algorithm of robust self-paced learning (R2SPL).

1: Input:

2: : source domain data;

3: : target domain data;

4: , : self-paced parameter; : regularization parameter

5: For each layer of self-paced learning:

6: Initiate parameter by utilizing the result of last self-paced learning layer.

7: Repeating until convergence

8: Update by Eq. (12);

9: Update by Eq. (1);

10: Update by Eq. (8);

11: Train the model based on and .