Webster, AshtonMachine learning techniques for defect and vulnerability detection have the potential to quickly direct developers' attention to software components with faulty implementations. Effective application of such defect prediction methods in practical software development environments requires transfer learning algorithms so that models built using existing projects can recognize defects as they emerge in a new project. Up until this study, comparing the efficacy of transfer learning algorithms was challenging because previous studies used differing data sets, baselines, and performance metrics. By providing open source implementations and baseline performance metrics for several transfer learning algorithms on two different data sets, our project offers software engineers the tools to objectively compare methods and readily identify top performing transfer learning algorithms in the domain of both vulnerability and defect prediction.en-USTechnical Report