DLFix: Context-based Code Transformation Learning for Automated Program Repair
Automated Program Repair (APR) is very useful in helping developers in the process of software development and maintenance. Despite recent advances in deep learning (DL), the DL-based APR approaches still have limitations in learning bug-fixing code changes and learning which context of the surrounding source code that certain bug-fixing changes should be made. These limitations lead to incorrect fixing locations or incorrect fixes. In this paper, we introduce DLFix, a two-tier DL model that treats APR as code transformation learning from the prior bug fixes and the surrounding code contexts of the fixes. The first layer is a tree-based RNN model that learns the contexts of bug fixes and its result is used as an additional weighting input for the second layer, which is designed to learn the bug-fixing code transformations.
We conducted several experiments to evaluate DLFix in two standard datasets Defects4J, and Bugs.jar, and in a newly built bug datasets with a total of +20K real-world bugs in eight projects. We have compared against a total of 13 state-of-the-art pattern-based APR tools. Our results show that DLFix improves over 11 of them, and is comparable and complementary to the top two pattern-based APR tools in which there are 7 and 11 unique bugs that they cannot detect, respectively, but we can. Importantly, DLFix is fully automated and data-driven, and does not require hard-coding of bug-fixing patterns as in those tools. We compared DLFix against 4 state-of-the-art deep learning based APR models. DLFix is able to fix 2.5 times more bugs than the best performing baseline.
Mon 5 Oct
|16:10 - 16:25|
Anil KoyuncuUniversity of Luxembourg, Luxembourg, Kui LiuUniversity of Luxembourg, Tegawendé F. BissyandéSnT, University of Luxembourg, Dongsun KimFuriosa.ai, Jacques KleinUniversity of Luxembourg, SnT, Martin MonperrusKTH Royal Institute of Technology, Yves Le TraonUniversity of LuxembourgPre-print
|16:25 - 16:45|
|16:45 - 17:00|
|17:00 - 17:20|
On the Efficiency of Test Suite based Program Repair: A Systematic Assessment of 16 Automated Repair Systems for Java ProgramsTechnical
Kui LiuUniversity of Luxembourg, Shangwen WangNational University of Defense Technology, Anil KoyuncuUniversity of Luxembourg, Luxembourg, Kisub KimUniversity of Luxembourg, SnT, Tegawendé F. BissyandéSnT, University of Luxembourg, Dongsun KimFuriosa.ai, Peng WuNational University of Defense Technology, Jacques KleinUniversity of Luxembourg, SnT, Xiaoguang MaoNational University of Defense Technology, Yves Le TraonUniversity of LuxembourgPre-print
|17:20 - 17:35|
|17:35 - 17:50|