期刊
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
卷 14, 期 -, 页码 3988-3999出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/JSTARS.2021.3069013
关键词
Training; Hyperspectral imaging; Feature extraction; Radio frequency; Deep learning; Convolutional neural networks; Stacking; Convolutional neural network (CNN); enhanced random feature subspace (ERFS); ensemble learning; hyperspectral image (HSI) classification; multiclass imbalance
类别
资金
- National Natural Science Foundation of China [61772397, 12005169]
- National Key R&D Program of China [2016YFE0200400]
- Science and Technology Innovation Team of Shaanxi Province [2019TD-002]
- Open Research Fund of Key Laboratory of Digital Earth Science [2019LDE005]
The article proposes a novel enhanced-random-feature-subspace-based ensemble CNN algorithm for multiclass imbalanced problem. By performing random oversampling and data enhancements based on random feature subspace, the algorithm combines random feature selection and CNN for HSI classification, showing better performance than traditional CNN, RF, and deep learning ensemble methods in experimental results on three public hyperspectral datasets.
Hyperspectral image (HSI) classification often faces the problem of multiclass imbalance, which is considered to be one of the major challenges in the field of remote sensing. In recent years, deep learning has been successfully applied to the HSI classification, a convolutional neural network (CNN) is one of the most representative of them. However, it is difficult to effectively improve the accuracy of minority classes under the problem of multiclass imbalance. In addition, ensemble learning has been successfully applied to solve multiclass imbalance, such as random forest (RF) This article proposes a novel enhanced-random-feature-subspace-based ensemble CNN algorithm for the multiclass imbalanced problem. The main idea is to perform random oversampling of training samples and multiple data enhancements based on random feature subspace, and then, construct an ensemble learning model combining random feature selection and CNN to the HSI classification. Experimental results on three public hyperspectral datasets show that the performance of the proposed method is better than the traditional CNN, RF, and deep learning ensemble methods.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据