Affiliations: School of Computer Sciences and Information Engineering, Xiamen University of Technology, Xiamen, China | School of Applied Mathematics, Xiamen University of Technology, Xiamen, China | School of Information Sciences and Engineering, Xiamen University, Xiamen, China
Note: [] Corresponding author. Ying Ma, School of Computer Sciences and Information Engineering, Xiamen University of Technology, Xiamen, China. Tel.: +86 0592 6291518; Fax: +86 0592 6291390; E-mail: [email protected]
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.
Keywords: Semi-supervised learning, defect prediction, random sampling, class imbalance, co-train
