Candidate: Thibaud Lutellier
Title: Machine Learning for Software Dependability
Date: October 7, 2020
Time: 9:00 AM
Place: REMOTE ATTENDANCE
Supervisor(s): Tan, Lin (Adjunct)
Abstract:
Dependability is an important quality of modern software but is challenging to achieve. Many software dependability techniques have been proposed to help developers improve software reliability and dependability such as defect prediction [76,88,220], bug detection [6, 17, 131], program repair [48, 114, 135, 188, 232, 234], test case prioritization [137, 221], or software architecture recovery [13, 41, 62, 100, 146, 212].
In this thesis we consider how machine learning (ML) and deep-learning (DL) can be used to enhanced software dependability through three examples in three different domains: automatic program repair, bug detection in electronic document readers, and software architecture recovery.
In the first work we propose a new G&V technique—CoCoNuT, which uses ensemble learning on the combination of convolutional neural networks (CNNs) and a new context-aware neural machine translation (NMT) architecture to automatically fix bugs in multiple programming languages. To better represent the context of a bug, we introduce a new context-aware NMT architecture that represents the buggy source code and its surrounding context separately. CoCoNuT uses CNNs instead of recurrent neural networks (RNNs), since CNN layers can be stacked to extract hierarchical features and better model source code at different granularity levels (e.g., statements and functions). In addition, CoCoNuT takes advantage of the randomness in hyperparameter tuning to build multiple models that fix different bugs and combines these models using ensemble learning to fix more bugs. CoCoNuT fixes 493 bugs, including 307 bugs that are fixed by none of the 27 techniques with which we compare.
In the second work, we present a study on the correctness of PDF documents and readers and propose an approach to detect and localize the source of such inconsistencies automatically. We evaluate our automatic approach on a large corpus of over 230K documents using 11 popular readers and our experiments have detected 30 unique bugs in these readers and files.
In the third work we compare software architecture recovery techniques to understand their effectiveness and applicability. Specifically, we study the impact of leveraging accurate symbol dependencies on the accuracy of architecture recovery techniques. In addition, we evaluate other factors of the input dependencies such as the level of granularity and the dynamic-bindings graph construction. The results of our evaluation of nine architecture recovery techniques and their variants suggest that (1) using accurate symbol dependencies has a major influence on recovery quality, and (2) more accurate recovery techniques are needed. Our results show that some of the studied architecture recovery techniques scale to very large systems, whereas others do not.