An improved semi-supervised learning method for software defect prediction

Abstract

This paper presents an improved semi-supervised learning approach for defect prediction involving class imbalanced and limited labeled data problem. This approach employs random under-sampling technique to resample the original training set and updating training set in each round for co-train style algorithm. It makes the defect predictor more practical for real applications, by combating these problems. In comparison with conventional machine learning approaches, our method has significant superior performance. Experimental results also show that with the proposed learning approach, it is possible to design better method to tackle the class imbalanced problem in semi-supervised learning.