Research
Software Defect Prediction Research
Innovations in Software Engineering Conference
February 22-24, 2024
Bengaluru, India
- Developed DEST, an ensemble-based self-training semi-supervised learning model for software defect prediction, and implemented bootstrap sampling strategies to generate labeled defect data, addressing data scarcity issues.
- Conducted analysis on 11 cross-software projects, achieving superior performance compared to other models.
- Key contributions include the development of DEST for large-scale labeled data generation and expertise in ensemble and semi-supervised learning using the R language.