Does Fixing Bug Increase Robustness in Deep Learning?
Deep Learning (DL) based systems are utilized vastly. Developers update the code to fix the bugs in the system. How these code fixing techniques impacts the robustness of these systems has not been clear. Does fixing code increase the robustness? Do they deteriorate the learning capability of the DL based systems? To answer these questions, we studied 321 Stack Overflow posts based on a published dataset. In this study, we built a classification scheme to analyze how bug-fixes change the robustness of the DL model and found that most of the bug-fixes can increase the robustness. We also found evidence of bug-fixing that decrease the robustness. Our preliminary result suggests that 12.5% and 2.4% of the bug-fixes in Stack Overflow posts caused the increase and the decrease of the robustness of DL models respectively.
Rangeet Pan is a Ph.D. student at Iowa State University. His research interests include program analysis, machine learning, and software engineering. He has published works at ESEC/FSE and ICSE.
Wed 8 JulDisplayed time zone: (UTC) Coordinated Universal Time change
02:10 - 03:00 | |||
02:10 50mPoster | Exploring the Relationship between Dockerfile Quality and Project Charateristics ACM Student Research Competition Yiwen Wu National University of Defense Technology | ||
02:10 50mPoster | Towards Automated Migration for Blockchain-based Decentralized Application ACM Student Research Competition Xiufeng Xu Peking University | ||
02:10 50mPoster | Stress Testing SMT Solvers via Type-aware Mutation ACM Student Research Competition Chengyu Zhang East China Normal University | ||
02:10 50mPoster | Does Fixing Bug Increase Robustness in Deep Learning? ACM Student Research Competition Rangeet Pan Iowa State University, USA | ||
02:10 50mPoster | Detection and Mitigation of JIT-Induced Side Channels ACM Student Research Competition Tegan Brennan University of California, Santa Barbara | ||
02:10 50mPoster | Uncertainty-Guided Testing and Robustness Enhancement for Deep Learning Systems ACM Student Research Competition Xiyue Zhang Peking University | ||
