Deep Learning-Based Fault Localization with Contextual Information

Abstract

SUMMARYFault localization is essential for solving the issue of software faults. Aiming at improving fault localization, this paper proposes a deep learning-based fault localization with contextual information. Specifically, our approach uses deep neural network to construct a suspiciousness evaluation model to evaluate the suspiciousness of a statement being faulty, and then leverages dynamic backward slicing to extract contextual information. The empirical results show that our approach significantly outperforms the state-of-the-art technique Dstar.

Locating Vulnerability in Binaries Using Deep Neural Networks

“…Zheng et al propose a fault localization method based on deep neural networks, which also aims to find out the executable statements containing bugs from source code [12]. Based on these researches, Zhang et al propose a deep learningbased fault localization method to evaluate the suspiciousness of statements, which leverages dynamic backward slicing to extract contextual information [32]. These researches focus on finding out suspicious executable statements of programs, but not for binary fault localization.…”

Improving deep‐learning‐based fault localization with resampling

“…Based on the methods proposed by Wong et al and advantages of deep learning methods, Zheng et al 12 construct a fault localization model using Multi‐Layer Perceptrons (MLP). Zhang et al 9 use dynamic slices to enhance fault localization effectiveness in the context of deep neural networks. Briand et al 36 propose a fault localization method using decision tree algorithm and construct those rules that classify test cases into various partitions.…”

“…The recent progress on deep learning shows its promising ability of learning useful models in various applications (e.g., image classification, object detection, and segmentation) and providing tremendous improvement in robustness and accuracy.8Some researchers have exploited the use of this learning ability to discuss and evaluate the potential of deep learning in fault localization. 9,10,12 Their research has shown that deep learning provides a new perspective for fault localization and can significantly improve localization effectiveness.…”

TFIDF-FL: Localizing Faults Using Term Frequency-Inverse Document Frequency and Deep Learning

“…Thus, some researchers have preliminarily used deep neural networks with multiple hidden layers to discuss and evaluate the potential of deep learning in fault localization [3], [4]. They found that with the capability of estimating complicated functions by learning a deep nonlinear network's structure and attaining distributed representation of input data, deep neural networks exhibit strong learning ability from sample data sets.…”