[Abstract]

Static analysis is one of the most widely used software engineering techniques in the software industry that predicts software behavior. For practical application, static analyzers need to be both precise and scalable, but developing such cost-effective analyzers is challenging. This is because practical static analyzers require high-quality analysis heuristics, but designing effective analysis heuristics is difficult and requires laborious tasks even for domain experts.

In this presentation, I will introduce my research on “Data-Driven Static Analysis” to address the problem by developing machine learning methodologies that automatically generate high-quality analysis heuristics from data. The learning framework has produced various heuristics that outperform the existing state-of-the-art. I will discuss how I have developed the learning framework and how I will improve the framework in the future.

[Biography]

Minseok Jeon is a postdoctoral researcher at Korea University. He received his B.S. from the Department of Computer Science and Engineering, College of Informatics at Korea University in 2017. He earned Ph.D. through an integrated M.S. & Ph.D. course from the same department in 2023. His research interest lies in program analysis. Especially, he has interests in developing machine learning methodologies for effective program analysis.