Semi-supervised generative adversarial nets with multiple generators for SAR image recognition

Fei Gao; Fei Ma; Jun Wang; Jinping Sun; Erfu Yang; Huiyu Zhou

doi:10.3390/s18082706

Semi-supervised generative adversarial nets with multiple generators for SAR image recognition

Fei Gao^*, Fei Ma, Jun Wang, Jinping Sun, Erfu Yang, Huiyu Zhou

^*Corresponding author for this work

School of Electronics, Electrical Engineering and Computer Science

Research output: Contribution to journal › Article › peer-review

24 Citations (Scopus)

60 Downloads (Pure)

Abstract

As an important model of deep learning, semi-supervised learning models are based on Generative Adversarial Nets (GANs) and have achieved a competitive performance on standard optical images. However, the training of GANs becomes unstable when they are applied to SAR images, which reduces the feature extraction capability of the discriminator in GANs. This paper presents a new semi-supervised GANs with Multiple generators and a classifier (MCGAN). This model improves the stability of training for SAR images by employing multiple generators. A multi-classifier is introduced to the new GANs to utilize the labeled images during the training of the GANs, which shares the low level layers with the discriminator. Then, the layers of the trained discriminator and the classifier construct the recognition network for SAR images after having been finely tuned using a small number of the labeled images. Experiments on the Moving and Stationary Target Acquisition and Recognition (MSTAR) databases show that the proposed recognition network achieves a better and more stable recognition performance than several traditional semi-supervised methods as well as other GANs-based semi-supervised methods.

Original language	English
Article number	2706
Journal	Sensors (Switzerland)
Volume	18
Issue number	8
DOIs	https://doi.org/10.3390/s18082706
Publication status	Published - 17 Aug 2018

Bibliographical note

Funding Information:
Funding: This research was funded by the National Natural Science Foundation of China, grant number 61771027, 61071139, 61471019, 61501011, and 61171122. E. Yang is supported in part under the RSE-NNSFC Joint Project (2017-2019), grant number 6161101383 with China University of Petroleum (Huadong). H. Zhou is supported by UK EPSRC, grant number EP/N508664/1, EP/R007187/1 and EP/N011074/1, and Royal Society-Newton Advanced Fellowship, grant number NA160342.

Publisher Copyright:
© 2018 by the authors. Licensee MDPI, Basel, Switzerland.

Copyright:
Copyright 2018 Elsevier B.V., All rights reserved.

Keywords

Deep learning
Generative adversarial networks (GANs)
Semi-supervised recognition
Synthetic aperture radar (SAR)

ASJC Scopus subject areas

Analytical Chemistry
Biochemistry
Atomic and Molecular Physics, and Optics
Instrumentation
Electrical and Electronic Engineering

Access to Document

10.3390/s18082706Licence: CC BY

Semi-supervised generative adversarial nets with multiple generators for SAR image recognition
Copyright 2018 the authors. This is an open access article published under a Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution and reproduction in any medium, provided the author and source are cited.
Final published version, 4.5 MBLicence: CC BY

Cite this

@article{be4d685137f44f24aacc8f2bf5179601,

title = "Semi-supervised generative adversarial nets with multiple generators for SAR image recognition",

abstract = "As an important model of deep learning, semi-supervised learning models are based on Generative Adversarial Nets (GANs) and have achieved a competitive performance on standard optical images. However, the training of GANs becomes unstable when they are applied to SAR images, which reduces the feature extraction capability of the discriminator in GANs. This paper presents a new semi-supervised GANs with Multiple generators and a classifier (MCGAN). This model improves the stability of training for SAR images by employing multiple generators. A multi-classifier is introduced to the new GANs to utilize the labeled images during the training of the GANs, which shares the low level layers with the discriminator. Then, the layers of the trained discriminator and the classifier construct the recognition network for SAR images after having been finely tuned using a small number of the labeled images. Experiments on the Moving and Stationary Target Acquisition and Recognition (MSTAR) databases show that the proposed recognition network achieves a better and more stable recognition performance than several traditional semi-supervised methods as well as other GANs-based semi-supervised methods.",

keywords = "Deep learning, Generative adversarial networks (GANs), Semi-supervised recognition, Synthetic aperture radar (SAR)",

author = "Fei Gao and Fei Ma and Jun Wang and Jinping Sun and Erfu Yang and Huiyu Zhou",

note = "Funding Information: Funding: This research was funded by the National Natural Science Foundation of China, grant number 61771027, 61071139, 61471019, 61501011, and 61171122. E. Yang is supported in part under the RSE-NNSFC Joint Project (2017-2019), grant number 6161101383 with China University of Petroleum (Huadong). H. Zhou is supported by UK EPSRC, grant number EP/N508664/1, EP/R007187/1 and EP/N011074/1, and Royal Society-Newton Advanced Fellowship, grant number NA160342. Publisher Copyright: {\textcopyright} 2018 by the authors. Licensee MDPI, Basel, Switzerland. Copyright: Copyright 2018 Elsevier B.V., All rights reserved.",

year = "2018",

month = aug,

day = "17",

doi = "10.3390/s18082706",

language = "English",

volume = "18",

journal = "Sensors (Switzerland)",

issn = "1424-8220",

publisher = "MDPI",

number = "8",

}

TY - JOUR

T1 - Semi-supervised generative adversarial nets with multiple generators for SAR image recognition

AU - Gao, Fei

AU - Ma, Fei

AU - Wang, Jun

AU - Sun, Jinping

AU - Yang, Erfu

AU - Zhou, Huiyu

N1 - Funding Information: Funding: This research was funded by the National Natural Science Foundation of China, grant number 61771027, 61071139, 61471019, 61501011, and 61171122. E. Yang is supported in part under the RSE-NNSFC Joint Project (2017-2019), grant number 6161101383 with China University of Petroleum (Huadong). H. Zhou is supported by UK EPSRC, grant number EP/N508664/1, EP/R007187/1 and EP/N011074/1, and Royal Society-Newton Advanced Fellowship, grant number NA160342. Publisher Copyright: © 2018 by the authors. Licensee MDPI, Basel, Switzerland. Copyright: Copyright 2018 Elsevier B.V., All rights reserved.

PY - 2018/8/17

Y1 - 2018/8/17

N2 - As an important model of deep learning, semi-supervised learning models are based on Generative Adversarial Nets (GANs) and have achieved a competitive performance on standard optical images. However, the training of GANs becomes unstable when they are applied to SAR images, which reduces the feature extraction capability of the discriminator in GANs. This paper presents a new semi-supervised GANs with Multiple generators and a classifier (MCGAN). This model improves the stability of training for SAR images by employing multiple generators. A multi-classifier is introduced to the new GANs to utilize the labeled images during the training of the GANs, which shares the low level layers with the discriminator. Then, the layers of the trained discriminator and the classifier construct the recognition network for SAR images after having been finely tuned using a small number of the labeled images. Experiments on the Moving and Stationary Target Acquisition and Recognition (MSTAR) databases show that the proposed recognition network achieves a better and more stable recognition performance than several traditional semi-supervised methods as well as other GANs-based semi-supervised methods.

AB - As an important model of deep learning, semi-supervised learning models are based on Generative Adversarial Nets (GANs) and have achieved a competitive performance on standard optical images. However, the training of GANs becomes unstable when they are applied to SAR images, which reduces the feature extraction capability of the discriminator in GANs. This paper presents a new semi-supervised GANs with Multiple generators and a classifier (MCGAN). This model improves the stability of training for SAR images by employing multiple generators. A multi-classifier is introduced to the new GANs to utilize the labeled images during the training of the GANs, which shares the low level layers with the discriminator. Then, the layers of the trained discriminator and the classifier construct the recognition network for SAR images after having been finely tuned using a small number of the labeled images. Experiments on the Moving and Stationary Target Acquisition and Recognition (MSTAR) databases show that the proposed recognition network achieves a better and more stable recognition performance than several traditional semi-supervised methods as well as other GANs-based semi-supervised methods.

KW - Deep learning

KW - Generative adversarial networks (GANs)

KW - Semi-supervised recognition

KW - Synthetic aperture radar (SAR)

U2 - 10.3390/s18082706

DO - 10.3390/s18082706

M3 - Article

C2 - 30126120

AN - SCOPUS:85052052861

SN - 1424-8220

VL - 18

JO - Sensors (Switzerland)

JF - Sensors (Switzerland)

IS - 8

M1 - 2706

ER -

Semi-supervised generative adversarial nets with multiple generators for SAR image recognition

Abstract

Bibliographical note

Keywords

ASJC Scopus subject areas

Access to Document

Fingerprint

Cite this