Balanced Adversarial Tight Matching for Cross-Project Defect Prediction
Algorithm 1
BATM training algorithm.
Input: Source project and label data target project data parameters of classifiers and and generators , and , learning rate , pseudo-label , output of the generator to the target project , maximum iteration .
Ouput:, and .
1: procedure BATM ()
2: Initialize the parameters , and randomly
3: for do
4: / Step 1 /
5: Compute the loss
6:
7:
8:
9: / Step 2 /
10: / Predicting pseudo-labels for target projects /