Quantitative Decision-Making for Automated Software Testing

Greybox fuzzing has emerged as the state-of-the-art technique for detecting software vulnerabilities. The increasing popularity and adoption of greybox fuzzing have highlighted the need for quantitative assessment of testing progress, both achieved and potential. In this study, we introduce statistical techniques that leverage existing testing data to gain insights and aid decision-making in the fuzzing process. We address the challenges that can impact the accuracy of estimation and prediction, providing tailored statistical tools. These tools are expected to serve as a valuable resource for practitioners, assisting them in evaluating testing progress and effectively anticipating future directions.