To improve software reliability, software defect prediction is used to find software bugs and prioritize testing efforts. Recently, some researchers introduced deep learning models, such as the deep belief network (DBN) and the state-of-the-art convolutional neural network (CNN), and used automatically generated features extracted from abstract syntax trees (ASTs) and deep learning models to improve defect prediction performance. However, the research on the CNN model failed to reveal clear conclusions due to its limited dataset size, insufficiently repeated experiments, and outdated baseline selection. To solve these problems, we built the PROMISE Source Code (PSC) dataset to enlarge the original dataset in the CNN research, which we named the Simplified PROMISE Source Code (SPSC) dataset. Then, we proposed an improved CNN model for within-project defect prediction (WPDP) and compared our results to existing CNN results and an empirical study. Our experiment was based on a 30-repetition holdout validation and a 10 * 10 cross-validation. Experimental results showed that our improved CNN model was comparable to the existing CNN model, and it outperformed the state-of-the-art machine learning models significantly for WPDP. Furthermore, we defined hyperparameter instability and examined the threat and opportunity it presents for deep learning models on defect prediction.

Deep learning-based software defect prediction has been popular these days. Recently, the publishing of the CodeBERT model has made it possible to perform many software engineering tasks. We propose various CodeBERT models targeting software defect prediction, including CodeBERT-NT, CodeBERT-PS, CodeBERT-PK, and CodeBERT-PT. We perform empirical studies using such models in cross-version and cross-project software defect prediction to investigate if using a neural language model like CodeBERT could improve prediction performance. We also investigate the effects of different prediction patterns in software defect prediction using CodeBERT models. The empirical results are further discussed.

No abstract

No abstract

No abstract