A novel active semisupervised convolutional neural network algorithm for SAR image recognition

Fei Gao, Zhenyu Yue, Jun Wang, Jinping Sun, Erfu Yang, Huiyu Zhou

Research output: Contribution to journalArticle

8 Citations (Scopus)

Abstract

Convolutional neural network (CNN) can be applied in synthetic aperture radar (SAR) object recognition for achieving good performance. However, it requires a large number of the labelled samples in its training phase, and therefore its performance could decrease dramatically when the labelled samples are insufficient. To solve this problem, in this paper, we present a novel active semisupervised CNN algorithm. First, the active learning is used to query the most informative and reliable samples in the unlabelled samples to extend the initial training dataset. Next, a semisupervised method is developed by adding a new regularization term into the loss function of CNN. As a result, the class probability information contained in the unlabelled samples can be maximally utilized. The experimental results on the MSTAR database demonstrate the effectiveness of the proposed algorithm despite the lack of the initial labelled samples.

LanguageEnglish
Article number3105053
Pages1-8
Number of pages8
JournalComputational Intelligence and Neuroscience
Volume2017
DOIs
Publication statusPublished - 1 Oct 2017

Fingerprint

Radar
Image recognition
Image Recognition
Synthetic Aperture
Network Algorithms
Synthetic aperture radar
Neural Networks
Neural networks
Problem-Based Learning
Object recognition
Databases
Active Learning
Object Recognition
Loss Function
Recognition (Psychology)
Regularization
Query
Decrease
Experimental Results
Term

Keywords

  • convolutional neural networks
  • synthetic aperture radar
  • machine learning
  • image recognition

Cite this

Gao, Fei ; Yue, Zhenyu ; Wang, Jun ; Sun, Jinping ; Yang, Erfu ; Zhou, Huiyu. / A novel active semisupervised convolutional neural network algorithm for SAR image recognition. In: Computational Intelligence and Neuroscience. 2017 ; Vol. 2017. pp. 1-8.
@article{783a5bfb93f94ac884b441173aab9e3b,
title = "A novel active semisupervised convolutional neural network algorithm for SAR image recognition",
abstract = "Convolutional neural network (CNN) can be applied in synthetic aperture radar (SAR) object recognition for achieving good performance. However, it requires a large number of the labelled samples in its training phase, and therefore its performance could decrease dramatically when the labelled samples are insufficient. To solve this problem, in this paper, we present a novel active semisupervised CNN algorithm. First, the active learning is used to query the most informative and reliable samples in the unlabelled samples to extend the initial training dataset. Next, a semisupervised method is developed by adding a new regularization term into the loss function of CNN. As a result, the class probability information contained in the unlabelled samples can be maximally utilized. The experimental results on the MSTAR database demonstrate the effectiveness of the proposed algorithm despite the lack of the initial labelled samples.",
keywords = "convolutional neural networks, synthetic aperture radar, machine learning, image recognition",
author = "Fei Gao and Zhenyu Yue and Jun Wang and Jinping Sun and Erfu Yang and Huiyu Zhou",
year = "2017",
month = "10",
day = "1",
doi = "10.1155/2017/3105053",
language = "English",
volume = "2017",
pages = "1--8",
journal = "Computational Intelligence and Neuroscience",
issn = "1687-5265",

}

A novel active semisupervised convolutional neural network algorithm for SAR image recognition. / Gao, Fei; Yue, Zhenyu; Wang, Jun; Sun, Jinping; Yang, Erfu; Zhou, Huiyu.

In: Computational Intelligence and Neuroscience, Vol. 2017, 3105053, 01.10.2017, p. 1-8.

Research output: Contribution to journalArticle

TY - JOUR

T1 - A novel active semisupervised convolutional neural network algorithm for SAR image recognition

AU - Gao, Fei

AU - Yue, Zhenyu

AU - Wang, Jun

AU - Sun, Jinping

AU - Yang, Erfu

AU - Zhou, Huiyu

PY - 2017/10/1

Y1 - 2017/10/1

N2 - Convolutional neural network (CNN) can be applied in synthetic aperture radar (SAR) object recognition for achieving good performance. However, it requires a large number of the labelled samples in its training phase, and therefore its performance could decrease dramatically when the labelled samples are insufficient. To solve this problem, in this paper, we present a novel active semisupervised CNN algorithm. First, the active learning is used to query the most informative and reliable samples in the unlabelled samples to extend the initial training dataset. Next, a semisupervised method is developed by adding a new regularization term into the loss function of CNN. As a result, the class probability information contained in the unlabelled samples can be maximally utilized. The experimental results on the MSTAR database demonstrate the effectiveness of the proposed algorithm despite the lack of the initial labelled samples.

AB - Convolutional neural network (CNN) can be applied in synthetic aperture radar (SAR) object recognition for achieving good performance. However, it requires a large number of the labelled samples in its training phase, and therefore its performance could decrease dramatically when the labelled samples are insufficient. To solve this problem, in this paper, we present a novel active semisupervised CNN algorithm. First, the active learning is used to query the most informative and reliable samples in the unlabelled samples to extend the initial training dataset. Next, a semisupervised method is developed by adding a new regularization term into the loss function of CNN. As a result, the class probability information contained in the unlabelled samples can be maximally utilized. The experimental results on the MSTAR database demonstrate the effectiveness of the proposed algorithm despite the lack of the initial labelled samples.

KW - convolutional neural networks

KW - synthetic aperture radar

KW - machine learning

KW - image recognition

UR - https://www.hindawi.com/journals/cin/contents/

U2 - 10.1155/2017/3105053

DO - 10.1155/2017/3105053

M3 - Article

VL - 2017

SP - 1

EP - 8

JO - Computational Intelligence and Neuroscience

T2 - Computational Intelligence and Neuroscience

JF - Computational Intelligence and Neuroscience

SN - 1687-5265

M1 - 3105053

ER -