FUSING DIFFERENT LEVELS OF DEEP FEATURES BY DEEP STACKED NEURAL NETWORK FOR HYPERSPECTRAL IMAGES

Shaohui Mei; Yanfu Chen; Jingyu Ji; Junhui Hou; Qian Du

doi:10.1109/IGARSS.2017.8127063

FUSING DIFFERENT LEVELS OF DEEP FEATURES BY DEEP STACKED NEURAL NETWORK FOR HYPERSPECTRAL IMAGES

Shaohui Mei
, Yanfu Chen
, Jingyu Ji
, Junhui Hou
, Qian Du

Department of Computer Science

Research output: Chapters, Conference Papers, Creative and Literary Works › RGC 32 - Refereed conference paper (with host publication) › peer-review

1 Citation (Scopus)

Abstract

Deep learning techniques have been demonstrated to be a powerful tool to learn features of images automatically. In this paper, a novel deep learning structure, i.e., deep stacked neural network (DSNN), is constructed to extract different levels of deep features of hyperspectral images. Specifically, convolutional neural network (CNN) is used as basic units in the proposed DSNN for feature extraction of hyperspectral images. Then, different levels of deep features are concatenated to form a novel fused feature for classification with a typical classifier, e.g., SVM. Experimental results on two benchmark hyperspectral datasets show that the fusion of features extracted in DSNN can produce higher classification accuracy than state-of-the-art deep learning based methods, indicating its effectiveness in feature learning.

Original language	English
Title of host publication	2017 IEEE International Geoscience & Remote Sensing Symposium
Subtitle of host publication	International Cooperation for Global Awareness, IGARSS 2017 - Proceedings
Publisher	IEEE
Pages	759-762
ISBN (Print)	9781509049516
DOIs	https://doi.org/10.1109/IGARSS.2017.8127063
Publication status	Published - Jul 2017
Event	37th IEEE International Geoscience and Remote Sensing Symposium (IGARSS 2017) - Fort Worth, United States Duration: 23 Jul 2017 → 28 Jul 2017 Conference number: 37th http://www.igarss2017.org/

Conference

Conference	37th IEEE International Geoscience and Remote Sensing Symposium (IGARSS 2017)
Abbreviated title	IGARSS 2017
Place	United States
City	Fort Worth
Period	23/07/17 → 28/07/17
Internet address	http://www.igarss2017.org/

Research Keywords

convolutional neural network
deep learning
deep stacked neural network
feature fusion
feature learning

Access Output

10.1109/IGARSS.2017.8127063

Other Access Options

Check CityUHK Library

Permanent Link

Right click to copy link

Cite this

Mei, S., Chen, Y., Ji, J., Hou, J., & Du, Q. (2017). FUSING DIFFERENT LEVELS OF DEEP FEATURES BY DEEP STACKED NEURAL NETWORK FOR HYPERSPECTRAL IMAGES. In 2017 IEEE International Geoscience & Remote Sensing Symposium: International Cooperation for Global Awareness, IGARSS 2017 - Proceedings (pp. 759-762). Article 8127063 IEEE. https://doi.org/10.1109/IGARSS.2017.8127063

@inproceedings{37c050045c684b9f818d25049f7af968,

title = "FUSING DIFFERENT LEVELS OF DEEP FEATURES BY DEEP STACKED NEURAL NETWORK FOR HYPERSPECTRAL IMAGES",

abstract = "Deep learning techniques have been demonstrated to be a powerful tool to learn features of images automatically. In this paper, a novel deep learning structure, i.e., deep stacked neural network (DSNN), is constructed to extract different levels of deep features of hyperspectral images. Specifically, convolutional neural network (CNN) is used as basic units in the proposed DSNN for feature extraction of hyperspectral images. Then, different levels of deep features are concatenated to form a novel fused feature for classification with a typical classifier, e.g., SVM. Experimental results on two benchmark hyperspectral datasets show that the fusion of features extracted in DSNN can produce higher classification accuracy than state-of-the-art deep learning based methods, indicating its effectiveness in feature learning.",

keywords = "convolutional neural network, deep learning, deep stacked neural network, feature fusion, feature learning",

author = "Shaohui Mei and Yanfu Chen and Jingyu Ji and Junhui Hou and Qian Du",

year = "2017",

month = jul,

doi = "10.1109/IGARSS.2017.8127063",

language = "English",

isbn = "9781509049516",

pages = "759--762",

booktitle = "2017 IEEE International Geoscience \& Remote Sensing Symposium",

publisher = "IEEE",

address = "United States",

note = "37th IEEE International Geoscience and Remote Sensing Symposium (IGARSS 2017), IGARSS 2017 ; Conference date: 23-07-2017 Through 28-07-2017",

url = "http://www.igarss2017.org/",

}

Mei, S, Chen, Y, Ji, J, Hou, J & Du, Q 2017, FUSING DIFFERENT LEVELS OF DEEP FEATURES BY DEEP STACKED NEURAL NETWORK FOR HYPERSPECTRAL IMAGES. in 2017 IEEE International Geoscience & Remote Sensing Symposium: International Cooperation for Global Awareness, IGARSS 2017 - Proceedings., 8127063, IEEE, pp. 759-762, 37th IEEE International Geoscience and Remote Sensing Symposium (IGARSS 2017), Fort Worth, Texas, United States, 23/07/17. https://doi.org/10.1109/IGARSS.2017.8127063

FUSING DIFFERENT LEVELS OF DEEP FEATURES BY DEEP STACKED NEURAL NETWORK FOR HYPERSPECTRAL IMAGES. / Mei, Shaohui; Chen, Yanfu; Ji, Jingyu et al.
2017 IEEE International Geoscience & Remote Sensing Symposium: International Cooperation for Global Awareness, IGARSS 2017 - Proceedings. IEEE, 2017. p. 759-762 8127063.

Research output: Chapters, Conference Papers, Creative and Literary Works › RGC 32 - Refereed conference paper (with host publication) › peer-review

TY - GEN

T1 - FUSING DIFFERENT LEVELS OF DEEP FEATURES BY DEEP STACKED NEURAL NETWORK FOR HYPERSPECTRAL IMAGES

AU - Mei, Shaohui

AU - Chen, Yanfu

AU - Ji, Jingyu

AU - Hou, Junhui

AU - Du, Qian

N1 - Conference code: 37th

PY - 2017/7

Y1 - 2017/7

N2 - Deep learning techniques have been demonstrated to be a powerful tool to learn features of images automatically. In this paper, a novel deep learning structure, i.e., deep stacked neural network (DSNN), is constructed to extract different levels of deep features of hyperspectral images. Specifically, convolutional neural network (CNN) is used as basic units in the proposed DSNN for feature extraction of hyperspectral images. Then, different levels of deep features are concatenated to form a novel fused feature for classification with a typical classifier, e.g., SVM. Experimental results on two benchmark hyperspectral datasets show that the fusion of features extracted in DSNN can produce higher classification accuracy than state-of-the-art deep learning based methods, indicating its effectiveness in feature learning.

AB - Deep learning techniques have been demonstrated to be a powerful tool to learn features of images automatically. In this paper, a novel deep learning structure, i.e., deep stacked neural network (DSNN), is constructed to extract different levels of deep features of hyperspectral images. Specifically, convolutional neural network (CNN) is used as basic units in the proposed DSNN for feature extraction of hyperspectral images. Then, different levels of deep features are concatenated to form a novel fused feature for classification with a typical classifier, e.g., SVM. Experimental results on two benchmark hyperspectral datasets show that the fusion of features extracted in DSNN can produce higher classification accuracy than state-of-the-art deep learning based methods, indicating its effectiveness in feature learning.

KW - convolutional neural network

KW - deep learning

KW - deep stacked neural network

KW - feature fusion

KW - feature learning

UR - https://www.scopus.com/pages/publications/85041847081

UR - https://www.scopus.com/record/pubmetrics.uri?eid=2-s2.0-85041847081&origin=recordpage

U2 - 10.1109/IGARSS.2017.8127063

DO - 10.1109/IGARSS.2017.8127063

M3 - RGC 32 - Refereed conference paper (with host publication)

SN - 9781509049516

SP - 759

EP - 762

BT - 2017 IEEE International Geoscience & Remote Sensing Symposium

PB - IEEE

T2 - 37th IEEE International Geoscience and Remote Sensing Symposium (IGARSS 2017)

Y2 - 23 July 2017 through 28 July 2017

ER -

FUSING DIFFERENT LEVELS OF DEEP FEATURES BY DEEP STACKED NEURAL NETWORK FOR HYPERSPECTRAL IMAGES

Abstract

Conference

Research Keywords

Access Output

Other Access Options

Permanent Link

Other Files and Links

Fingerprint

Cite this