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 inter-projets (CPDP) est une solution réalisable pour créer un modèle de prédiction précis sans suffisamment de données historiques. Bien que les méthodes existantes pour le CPDP qui utilisent uniquement des données étiquetées pour construire le modèle de prédiction obtiennent d'excellents résultats, il reste encore beaucoup à faire pour améliorer les performances de prédiction. Dans cet article, nous proposons une approche d'apprentissage semi-supervisé des caractéristiques discriminantes (SSDFL) pour le CPDP. SSDFL transfère d'abord les connaissances sur les données source et cible dans l'espace commun en utilisant un réseau neuronal entièrement connecté pour exploiter les similitudes potentielles entre les données source et cible. Ensuite, nous réduisons les différences de distributions marginales et de distributions conditionnelles entre les données source et cible cartographiées. Nous introduisons également la caractéristique discriminante d'apprentissage pour utiliser pleinement les informations d'étiquette, c'est-à-dire que les instances de la même classe sont proches les unes des autres et que les instances de différentes classes sont éloignées les unes des autres. Des expériences approfondies sont menées sur 10 projets issus des ensembles de données de l'AEEEM et de la NASA, et les résultats expérimentaux indiquent que notre approche obtient de meilleures performances de prédiction que les lignes de base.
Danlei XING
Nanjing University of Posts and Telecommunications
Fei WU
Nanjing University of Posts and Telecommunications
Ying SUN
Nanjing University of Posts and Telecommunications
Xiao-Yuan JING
Nanjing University of Posts and Telecommunications
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
Danlei XING, Fei WU, Ying SUN, Xiao-Yuan JING, "Cross-Project Defect Prediction via Semi-Supervised Discriminative Feature Learning" in IEICE TRANSACTIONS on Information,
vol. E103-D, no. 10, pp. 2237-2240, October 2020, doi: 10.1587/transinf.2020EDL8044.
Abstract: Cross-project defect prediction (CPDP) is a feasible solution to build an accurate prediction model without enough historical data. Although existing methods for CPDP that use only labeled data to build the prediction model achieve great results, there are much room left to further improve on prediction performance. In this paper we propose a Semi-Supervised Discriminative Feature Learning (SSDFL) approach for CPDP. SSDFL first transfers knowledge of source and target data into the common space by using a fully-connected neural network to mine potential similarities of source and target data. Next, we reduce the differences of both marginal distributions and conditional distributions between mapped source and target data. We also introduce the discriminative feature learning to make full use of label information, which is that the instances from the same class are close to each other and the instances from different classes are distant from each other. Extensive experiments are conducted on 10 projects from AEEEM and NASA datasets, and the experimental results indicate that our approach obtains better prediction performance than baselines.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2020EDL8044/_p
Copier
@ARTICLE{e103-d_10_2237,
author={Danlei XING, Fei WU, Ying SUN, Xiao-Yuan JING, },
journal={IEICE TRANSACTIONS on Information},
title={Cross-Project Defect Prediction via Semi-Supervised Discriminative Feature Learning},
year={2020},
volume={E103-D},
number={10},
pages={2237-2240},
abstract={Cross-project defect prediction (CPDP) is a feasible solution to build an accurate prediction model without enough historical data. Although existing methods for CPDP that use only labeled data to build the prediction model achieve great results, there are much room left to further improve on prediction performance. In this paper we propose a Semi-Supervised Discriminative Feature Learning (SSDFL) approach for CPDP. SSDFL first transfers knowledge of source and target data into the common space by using a fully-connected neural network to mine potential similarities of source and target data. Next, we reduce the differences of both marginal distributions and conditional distributions between mapped source and target data. We also introduce the discriminative feature learning to make full use of label information, which is that the instances from the same class are close to each other and the instances from different classes are distant from each other. Extensive experiments are conducted on 10 projects from AEEEM and NASA datasets, and the experimental results indicate that our approach obtains better prediction performance than baselines.},
keywords={},
doi={10.1587/transinf.2020EDL8044},
ISSN={1745-1361},
month={October},}
Copier
TY - JOUR
TI - Cross-Project Defect Prediction via Semi-Supervised Discriminative Feature Learning
T2 - IEICE TRANSACTIONS on Information
SP - 2237
EP - 2240
AU - Danlei XING
AU - Fei WU
AU - Ying SUN
AU - Xiao-Yuan JING
PY - 2020
DO - 10.1587/transinf.2020EDL8044
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E103-D
IS - 10
JA - IEICE TRANSACTIONS on Information
Y1 - October 2020
AB - Cross-project defect prediction (CPDP) is a feasible solution to build an accurate prediction model without enough historical data. Although existing methods for CPDP that use only labeled data to build the prediction model achieve great results, there are much room left to further improve on prediction performance. In this paper we propose a Semi-Supervised Discriminative Feature Learning (SSDFL) approach for CPDP. SSDFL first transfers knowledge of source and target data into the common space by using a fully-connected neural network to mine potential similarities of source and target data. Next, we reduce the differences of both marginal distributions and conditional distributions between mapped source and target data. We also introduce the discriminative feature learning to make full use of label information, which is that the instances from the same class are close to each other and the instances from different classes are distant from each other. Extensive experiments are conducted on 10 projects from AEEEM and NASA datasets, and the experimental results indicate that our approach obtains better prediction performance than baselines.
ER -