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
Durante os últimos anos, a aprendizagem profunda alcançou excelentes resultados em reconhecimento de imagem, processamento de voz e outras áreas de pesquisa, o que desencadeou um novo surto de pesquisa e aplicação. Defeitos internos e ataques maliciosos externos podem ameaçar a operação segura e confiável de um sistema de aprendizagem profunda e até causar consequências insuportáveis. A tecnologia de teste de sistemas de aprendizagem profunda ainda está em sua infância. A tecnologia tradicional de teste de software não é aplicável para testar sistemas de aprendizagem profunda. Além disso, as características do aprendizado profundo, como cenários de aplicação complexos, a alta dimensionalidade dos dados de entrada e a fraca interpretabilidade da lógica de operação, trazem novos desafios ao trabalho de teste. Este artigo enfoca o problema de geração de casos de teste e aponta que exemplos adversários podem ser usados como casos de teste. Em seguida, o artigo propõe o MTGAN, que é um framework para gerar casos de teste para classificadores de imagens de aprendizagem profunda baseados em Rede Adversarial Generativa. Finalmente, este artigo avalia a eficácia do MTGAN.
Erhu LIU
Army Engineering University of PLA,94973 Troop, Hangzhou
Song HUANG
Army Engineering University of PLA
Cheng ZONG
Army Engineering University of PLA
Changyou ZHENG
Army Engineering University of PLA
Yongming YAO
Army Engineering University of PLA
Jing ZHU
Army Engineering University of PLA
Shiqi TANG
Army Engineering University of PLA
Yanqiu WANG
Baopo Technology Co. Ltd.
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.
Copiar
Erhu LIU, Song HUANG, Cheng ZONG, Changyou ZHENG, Yongming YAO, Jing ZHU, Shiqi TANG, Yanqiu WANG, "MTGAN: Extending Test Case set for Deep Learning Image Classifier" in IEICE TRANSACTIONS on Information,
vol. E104-D, no. 5, pp. 709-722, May 2021, doi: 10.1587/transinf.2020EDP7162.
Abstract: During the recent several years, deep learning has achieved excellent results in image recognition, voice processing, and other research areas, which has set off a new upsurge of research and application. Internal defects and external malicious attacks may threaten the safe and reliable operation of a deep learning system and even cause unbearable consequences. The technology of testing deep learning systems is still in its infancy. Traditional software testing technology is not applicable to test deep learning systems. In addition, the characteristics of deep learning such as complex application scenarios, the high dimensionality of input data, and poor interpretability of operation logic bring new challenges to the testing work. This paper focuses on the problem of test case generation and points out that adversarial examples can be used as test cases. Then the paper proposes MTGAN which is a framework to generate test cases for deep learning image classifiers based on Generative Adversarial Network. Finally, this paper evaluates the effectiveness of MTGAN.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2020EDP7162/_p
Copiar
@ARTICLE{e104-d_5_709,
author={Erhu LIU, Song HUANG, Cheng ZONG, Changyou ZHENG, Yongming YAO, Jing ZHU, Shiqi TANG, Yanqiu WANG, },
journal={IEICE TRANSACTIONS on Information},
title={MTGAN: Extending Test Case set for Deep Learning Image Classifier},
year={2021},
volume={E104-D},
number={5},
pages={709-722},
abstract={During the recent several years, deep learning has achieved excellent results in image recognition, voice processing, and other research areas, which has set off a new upsurge of research and application. Internal defects and external malicious attacks may threaten the safe and reliable operation of a deep learning system and even cause unbearable consequences. The technology of testing deep learning systems is still in its infancy. Traditional software testing technology is not applicable to test deep learning systems. In addition, the characteristics of deep learning such as complex application scenarios, the high dimensionality of input data, and poor interpretability of operation logic bring new challenges to the testing work. This paper focuses on the problem of test case generation and points out that adversarial examples can be used as test cases. Then the paper proposes MTGAN which is a framework to generate test cases for deep learning image classifiers based on Generative Adversarial Network. Finally, this paper evaluates the effectiveness of MTGAN.},
keywords={},
doi={10.1587/transinf.2020EDP7162},
ISSN={1745-1361},
month={May},}
Copiar
TY - JOUR
TI - MTGAN: Extending Test Case set for Deep Learning Image Classifier
T2 - IEICE TRANSACTIONS on Information
SP - 709
EP - 722
AU - Erhu LIU
AU - Song HUANG
AU - Cheng ZONG
AU - Changyou ZHENG
AU - Yongming YAO
AU - Jing ZHU
AU - Shiqi TANG
AU - Yanqiu WANG
PY - 2021
DO - 10.1587/transinf.2020EDP7162
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E104-D
IS - 5
JA - IEICE TRANSACTIONS on Information
Y1 - May 2021
AB - During the recent several years, deep learning has achieved excellent results in image recognition, voice processing, and other research areas, which has set off a new upsurge of research and application. Internal defects and external malicious attacks may threaten the safe and reliable operation of a deep learning system and even cause unbearable consequences. The technology of testing deep learning systems is still in its infancy. Traditional software testing technology is not applicable to test deep learning systems. In addition, the characteristics of deep learning such as complex application scenarios, the high dimensionality of input data, and poor interpretability of operation logic bring new challenges to the testing work. This paper focuses on the problem of test case generation and points out that adversarial examples can be used as test cases. Then the paper proposes MTGAN which is a framework to generate test cases for deep learning image classifiers based on Generative Adversarial Network. Finally, this paper evaluates the effectiveness of MTGAN.
ER -