The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. ex. Some numerals are expressed as "XNUMX".
Copyrights notice
The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. Copyrights notice
La prédiction des défauts logiciels (SDP) joue un rôle essentiel dans l’allocation raisonnable des ressources de test et dans la garantie de la qualité des logiciels. Lorsqu'il n'y a pas suffisamment de modules historiques étiquetés, de nombreuses méthodes SDP semi-supervisées ont été proposées, et ces méthodes utilisent simultanément des modules étiquetés limités et de nombreux modules non étiquetés. Néanmoins, la plupart d’entre eux utilisent des fonctionnalités traditionnelles plutôt que les puissantes représentations de fonctionnalités profondes. En outre, le coût d'une mauvaise classification des modules défectueux est plus élevé que celui des modules sans défaut, et le nombre de modules défectueux pour la formation est faible. En tenant compte des problèmes ci-dessus, nous proposons un réseau à échelles clairsemé et sensible aux coûts (CSLN) pour SDP. Nous introduisons d’abord le réseau en échelle semi-supervisé pour extraire les représentations de fonctionnalités profondes. En outre, nous introduisons l'apprentissage sensible aux coûts pour définir différents coûts de classification erronée pour les instances sujettes aux défauts et celles sujettes aux défauts afin d'atténuer le problème de déséquilibre des classes. Une contrainte clairsemée est ajoutée sur les nœuds cachés dans le réseau à relais lorsque le nombre de nœuds cachés est grand, ce qui permet au modèle de trouver des structures robustes des données. Des expériences approfondies sur l'ensemble de données AEEEM montrent que le CSLN surpasse plusieurs méthodes SDP semi-supervisées de pointe.
Jing SUN
Nanjing University of Posts and Telecommunications (NJUPT)
Yi-mu JI
Nanjing University of Posts and Telecommunications (NJUPT)
Shangdong LIU
Nanjing University of Posts and Telecommunications (NJUPT)
Fei WU
NJUPT
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copier
Jing SUN, Yi-mu JI, Shangdong LIU, Fei WU, "Cost-Sensitive and Sparse Ladder Network for Software Defect Prediction" in IEICE TRANSACTIONS on Information,
vol. E103-D, no. 5, pp. 1177-1180, May 2020, doi: 10.1587/transinf.2019EDL8198.
Abstract: Software defect prediction (SDP) plays a vital role in allocating testing resources reasonably and ensuring software quality. When there are not enough labeled historical modules, considerable semi-supervised SDP methods have been proposed, and these methods utilize limited labeled modules and abundant unlabeled modules simultaneously. Nevertheless, most of them make use of traditional features rather than the powerful deep feature representations. Besides, the cost of the misclassification of the defective modules is higher than that of defect-free ones, and the number of the defective modules for training is small. Taking the above issues into account, we propose a cost-sensitive and sparse ladder network (CSLN) for SDP. We firstly introduce the semi-supervised ladder network to extract the deep feature representations. Besides, we introduce the cost-sensitive learning to set different misclassification costs for defective-prone and defect-free-prone instances to alleviate the class imbalance problem. A sparse constraint is added on the hidden nodes in ladder network when the number of hidden nodes is large, which enables the model to find robust structures of the data. Extensive experiments on the AEEEM dataset show that the CSLN outperforms several state-of-the-art semi-supervised SDP methods.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2019EDL8198/_p
Copier
@ARTICLE{e103-d_5_1177,
author={Jing SUN, Yi-mu JI, Shangdong LIU, Fei WU, },
journal={IEICE TRANSACTIONS on Information},
title={Cost-Sensitive and Sparse Ladder Network for Software Defect Prediction},
year={2020},
volume={E103-D},
number={5},
pages={1177-1180},
abstract={Software defect prediction (SDP) plays a vital role in allocating testing resources reasonably and ensuring software quality. When there are not enough labeled historical modules, considerable semi-supervised SDP methods have been proposed, and these methods utilize limited labeled modules and abundant unlabeled modules simultaneously. Nevertheless, most of them make use of traditional features rather than the powerful deep feature representations. Besides, the cost of the misclassification of the defective modules is higher than that of defect-free ones, and the number of the defective modules for training is small. Taking the above issues into account, we propose a cost-sensitive and sparse ladder network (CSLN) for SDP. We firstly introduce the semi-supervised ladder network to extract the deep feature representations. Besides, we introduce the cost-sensitive learning to set different misclassification costs for defective-prone and defect-free-prone instances to alleviate the class imbalance problem. A sparse constraint is added on the hidden nodes in ladder network when the number of hidden nodes is large, which enables the model to find robust structures of the data. Extensive experiments on the AEEEM dataset show that the CSLN outperforms several state-of-the-art semi-supervised SDP methods.},
keywords={},
doi={10.1587/transinf.2019EDL8198},
ISSN={1745-1361},
month={May},}
Copier
TY - JOUR
TI - Cost-Sensitive and Sparse Ladder Network for Software Defect Prediction
T2 - IEICE TRANSACTIONS on Information
SP - 1177
EP - 1180
AU - Jing SUN
AU - Yi-mu JI
AU - Shangdong LIU
AU - Fei WU
PY - 2020
DO - 10.1587/transinf.2019EDL8198
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E103-D
IS - 5
JA - IEICE TRANSACTIONS on Information
Y1 - May 2020
AB - Software defect prediction (SDP) plays a vital role in allocating testing resources reasonably and ensuring software quality. When there are not enough labeled historical modules, considerable semi-supervised SDP methods have been proposed, and these methods utilize limited labeled modules and abundant unlabeled modules simultaneously. Nevertheless, most of them make use of traditional features rather than the powerful deep feature representations. Besides, the cost of the misclassification of the defective modules is higher than that of defect-free ones, and the number of the defective modules for training is small. Taking the above issues into account, we propose a cost-sensitive and sparse ladder network (CSLN) for SDP. We firstly introduce the semi-supervised ladder network to extract the deep feature representations. Besides, we introduce the cost-sensitive learning to set different misclassification costs for defective-prone and defect-free-prone instances to alleviate the class imbalance problem. A sparse constraint is added on the hidden nodes in ladder network when the number of hidden nodes is large, which enables the model to find robust structures of the data. Extensive experiments on the AEEEM dataset show that the CSLN outperforms several state-of-the-art semi-supervised SDP methods.
ER -