공지사항

Date: 2024. January 5th
Time: 3 pm
Location: 104 E101

Seminar Title: Statistical Program Analysis
Speaker: Dr. Seongmin Lee, Researcher at Max Planck Institute

Abstract:

When facing a problem that is too complex for the analytical method, especially when it is unmanageable to compute a quantity exactly, a sampling-based statistical method can be used to overcome the limitation. It is well-known that Monte Carlo methods have been successfully applied to numerous problems across various fields, including natural sciences and engineering, where the solution is intractable for analytic computation.

In this talk, we share the glowing vision of beneficial applications of statistical methods in the field of program analysis. We introduce two recent successes in applying statistical methods to solve enigmatic problems in program analysis. The first solves the quantitative reachability analysis problem, which measures the probability of reaching a target program state given the workload of the software. The second extrapolates the progress of the Greybox fuzzing campaign, which lets us estimate how much more code coverage can be achieved by fuzzing. Both of these problems are intractable/unscalable for the analytical method. We show that the statistical method can be used to solve these problems efficiently and effectively.

Bio:

– Postdoctoral Researcher (Sep. 2022 – Present): Max Planck Institute for Security and Privacy (MPI-SP), Software Security Research Group

– Doctor of Philosophy, School of Computing (Sep. 2016 – Aug. 2022): KAIST

– Bachelor of Science, School of Computing and Bachelor of Science, Department of Mathematical Sciences (Feb. 2012 – Aug. 2016): KAIST

– Awards and honors

– PhD Dissertation Award, School of Computing, KAIST, 2022

– 2021 Naver Ph.D. Fellowship Award