Title: Automated software testing via extensive program analysis
Software faults (a.k.a., bugs) are prevalent in modern software systems, incurring the great loss of properties and even lives. Automated software testing can help verify whether the system is working in the same way as expected and ensure that the software is bug-free. Fault localization is one useful software testing technique to localize faults when certain test cases fail. In this talk, I will introduce a series of fault localization techniques which use various knowledge such as deep learning and the feedback information from program repair, i.e., DeepFL and ProFL. Especially, DeepFL can localize 213 out of 395 real-world faults, achieving the best result among all fault localization techniques back then. ProFL further improves DeepFL and has been successfully deployed in Alibaba Group. According to feedback from developers in Alibaba, ProFL can help save hours for manual debugging.
Xia Li is a Ph.D. candidate major in Computer Science at the University of Texas at Dallas. His research interests lie in Software Engineering, focusing on software testing and debugging. In particular, he seeks to improve automated fault localization (FL) via applying program analysis and machine learning/deep learning techniques in tandem. He has published some research papers in top-tier conferences and journals, such as ISSTA, OOPSLA and TSE. Especially, the most recent FL technique ProFL has been deployed in the industry (Alibaba Group), helping developers debug more efficiently.
Monday, February 17 at 11:15am to 12:05pm
University Hall, 2009
1402 10th Ave S, Birmingham, Alabama 35294