| 日時(Date) |
Oct 27th, 2026 4th period (15:00-–16:00) |
|---|---|
| 場所(Location) | パナソニック デジタル大講義室 (L1), Panasonic Digital Lecture Hall (L1) |
| 司会(Chair) | Zhenlan Ji sensei |
| 講演者(Presenter) | Dr. Wai Kin WONG, Hong Kong University of Science and Technology, Hong Kong |
| 題目(Title) | Understanding Software Internals for Automated Testing |
| 概要(Abstract) | Automated testing approach are often designed to be generic, such that they can be applied to a wide range of programs. However, this generality can limit their effectiveness, as they may overlook important internal behaviors that are specific to certain software systems. My research develops techniques for recovering and examining software internals so that automated tools can explore behavior that conventional methods overlook. In this talk, I will present automated techniques that recover and use information about software internals to guide analysis and testing. DecLLM repairs decompiler output into recompilable code, enabling source-based tools to analyze software available only as binaries. In JavaScript engines, StateLens automatically identifies internal runtime states that distinguish executions with identical code coverage, allowing fuzzers to explore behaviors that coverage alone would miss. At the software-hardware boundary, GPUDiff uses knowledge of GPU architecture to generate tests that stress compiler optimizations. By comparing execution times across compiler versions, it exposes performance regressions even when program outputs remain correct. I will conclude by discussing future directions for using program understanding to build more scalable and effective testing frameworks. |
| 講演言語(Language) | English |
| 講演者紹介(Introduction of Lecturer) | Wai Kin Wong is a postdoctoral researcher in the Department of Computer Science and Engineering at the Hong Kong University of Science and Technology. He received his Ph.D. from the same department under the supervision of Prof. Shuai Wang. His research lies at the intersection of software security and reverse engineering, with a focus on improving the dependability of software systems. His work has been published in top-tier venues, including ASPLOS, EuroSys, SOSP, and ICSE. |