Improving random forest with ensemble of features and semi-supervised feature extraction

Junshi Xia, Wenzhi Liao, Jocelyn Chanussot, Peijun Du, Guanghan Song, Wilfried Philips, Alejandro Frery

Research output: Contribution to journalArticle

18 Citations (Scopus)

Abstract

In this letter, we propose a novel approach for improving Random Forest (RF) in hyperspectral image classification. The proposed approach combines the ensemble of features and the semisupervised feature extraction (SSFE) technique. The main contribution of our approach is to construct an ensemble of RF classifiers. In this way, the feature space is divided into several disjoint feature subspaces. Then, the feature subspaces induced by the SSFE technique are used as the input space to an RF classifier. This method is compared with a regular RF and an RF with the reduced features by the SSFE on two real hyperspectral data sets, showing an improved performance in ill-posed, poor-posed, and well-posed conditions. An additional study shows that the proposed method is less sensitive to the parameters.
LanguageEnglish
Pages1471-1475
Number of pages5
JournalIEEE Geoscience and Remote Sensing Letters
Volume12
Issue number7
DOIs
Publication statusPublished - 16 Mar 2015

Fingerprint

Feature extraction
Classifiers
Image classification
image classification
method
parameter

Keywords

  • ensemble learning
  • semi-supervised feature extraction
  • classification
  • hyperspectral image
  • random forest
  • image classification
  • hyperspectral data
  • selection
  • accuracy
  • geophysical image processing

Cite this

Xia, Junshi ; Liao, Wenzhi ; Chanussot, Jocelyn ; Du, Peijun ; Song, Guanghan ; Philips, Wilfried ; Frery, Alejandro. / Improving random forest with ensemble of features and semi-supervised feature extraction. In: IEEE Geoscience and Remote Sensing Letters. 2015 ; Vol. 12, No. 7. pp. 1471-1475.
@article{bf99562d95e54bdb8e81c4c651602ac8,
title = "Improving random forest with ensemble of features and semi-supervised feature extraction",
abstract = "In this letter, we propose a novel approach for improving Random Forest (RF) in hyperspectral image classification. The proposed approach combines the ensemble of features and the semisupervised feature extraction (SSFE) technique. The main contribution of our approach is to construct an ensemble of RF classifiers. In this way, the feature space is divided into several disjoint feature subspaces. Then, the feature subspaces induced by the SSFE technique are used as the input space to an RF classifier. This method is compared with a regular RF and an RF with the reduced features by the SSFE on two real hyperspectral data sets, showing an improved performance in ill-posed, poor-posed, and well-posed conditions. An additional study shows that the proposed method is less sensitive to the parameters.",
keywords = "ensemble learning, semi-supervised feature extraction, classification, hyperspectral image, random forest, image classification, hyperspectral data, selection, accuracy, geophysical image processing",
author = "Junshi Xia and Wenzhi Liao and Jocelyn Chanussot and Peijun Du and Guanghan Song and Wilfried Philips and Alejandro Frery",
year = "2015",
month = "3",
day = "16",
doi = "10.1109/LGRS.2015.2409112",
language = "English",
volume = "12",
pages = "1471--1475",
journal = "IEEE Geoscience and Remote Sensing Letters",
issn = "1545-598X",
publisher = "IEEE",
number = "7",

}

Improving random forest with ensemble of features and semi-supervised feature extraction. / Xia, Junshi; Liao, Wenzhi; Chanussot, Jocelyn; Du, Peijun; Song, Guanghan; Philips, Wilfried; Frery, Alejandro.

In: IEEE Geoscience and Remote Sensing Letters, Vol. 12, No. 7, 16.03.2015, p. 1471-1475.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Improving random forest with ensemble of features and semi-supervised feature extraction

AU - Xia, Junshi

AU - Liao, Wenzhi

AU - Chanussot, Jocelyn

AU - Du, Peijun

AU - Song, Guanghan

AU - Philips, Wilfried

AU - Frery, Alejandro

PY - 2015/3/16

Y1 - 2015/3/16

N2 - In this letter, we propose a novel approach for improving Random Forest (RF) in hyperspectral image classification. The proposed approach combines the ensemble of features and the semisupervised feature extraction (SSFE) technique. The main contribution of our approach is to construct an ensemble of RF classifiers. In this way, the feature space is divided into several disjoint feature subspaces. Then, the feature subspaces induced by the SSFE technique are used as the input space to an RF classifier. This method is compared with a regular RF and an RF with the reduced features by the SSFE on two real hyperspectral data sets, showing an improved performance in ill-posed, poor-posed, and well-posed conditions. An additional study shows that the proposed method is less sensitive to the parameters.

AB - In this letter, we propose a novel approach for improving Random Forest (RF) in hyperspectral image classification. The proposed approach combines the ensemble of features and the semisupervised feature extraction (SSFE) technique. The main contribution of our approach is to construct an ensemble of RF classifiers. In this way, the feature space is divided into several disjoint feature subspaces. Then, the feature subspaces induced by the SSFE technique are used as the input space to an RF classifier. This method is compared with a regular RF and an RF with the reduced features by the SSFE on two real hyperspectral data sets, showing an improved performance in ill-posed, poor-posed, and well-posed conditions. An additional study shows that the proposed method is less sensitive to the parameters.

KW - ensemble learning

KW - semi-supervised feature extraction

KW - classification

KW - hyperspectral image

KW - random forest

KW - image classification

KW - hyperspectral data

KW - selection

KW - accuracy

KW - geophysical image processing

UR - http://hdl.handle.net/1854/LU-5847840

U2 - 10.1109/LGRS.2015.2409112

DO - 10.1109/LGRS.2015.2409112

M3 - Article

VL - 12

SP - 1471

EP - 1475

JO - IEEE Geoscience and Remote Sensing Letters

T2 - IEEE Geoscience and Remote Sensing Letters

JF - IEEE Geoscience and Remote Sensing Letters

SN - 1545-598X

IS - 7

ER -