Graph-Based Deep Multitask Few-Shot Learning for Hyperspectral Image Classification

被引:2
|
作者
Li, Na [1 ]
Zhou, Deyun [1 ]
Shi, Jiao [1 ]
Zheng, Xiaolong [1 ]
Wu, Tao [1 ]
Yang, Zhen [2 ]
机构
[1] Northwestern Polytech Univ, Sch Elect & Informat, 127 West Youyi Rd, Xian 710072, Peoples R China
[2] Beihang Univ, Sch Automat Sci & Elect Engn, Beijing 100191, Peoples R China
基金
中国国家自然科学基金; 中国博士后科学基金;
关键词
few-shot learning; graph; few labeled samples; hyperspectral images; semi-supervised; classification; DIMENSIONALITY REDUCTION; NETWORK;
D O I
10.3390/rs14092246
中图分类号
X [环境科学、安全科学];
学科分类号
08 ; 0830 ;
摘要
Although the deep neural network (DNN) has shown a powerful ability in hyperspectral image (HSI) classification, its learning requires a large number of labeled training samples; otherwise, it is prone to over-fitting and has a poor classification performance. However, this requirement is impractical for HSIs due to the difficulty in obtaining class labels. To make DNNs suitable for HSI classification with few labeled samples, we propose a graph-based deep multitask few-shot learning (GDMFSL) framework that learns the intrinsic relationships among all samples (labeled and unlabeled) of HSIs with the assistance of graph information to alleviate the over-fitting caused by few labeled training samples. Firstly, a semi-supervised graph is constructed to generate graph information. Secondly, a deep multitask network (DMN) is designed, which contains two subnetworks (tasks): a classifier subnetwork for learning class information from labeled samples and a Siamese subnetwork for learning sample relationships from the semi-supervised graph. To effectively learn graph information, a loss function suitable for the Siamese subnetwork is designed that shortens (and expands) the distance between the target sample and its nearest (and farthest) neighbors. Finally, since the number of training samples of the two subnetworks is severely imbalanced, a multitask few-shot learning strategy is designed to make two subnetworks converge simultaneously. Experimental results on the Indian Pines, University of Pavia and Salinas datasets demonstrate that GDMFSL achieves a better classification performance relative to existing competitors in few-shot settings. In particular, when only five labels per class are involved in training, the classification accuracy of GDMFSL on the three datasets reaches 87.58%, 86.42% and 98.85%, respectively.
引用
收藏
页数:27
相关论文
共 50 条
  • [1] Graph-Based Domain Adaptation Few-Shot Learning for Hyperspectral Image Classification
    Xu, Yanbing
    Zhang, Yanmei
    Yue, Tingxuan
    Yu, Chengcheng
    Li, Huan
    [J]. REMOTE SENSING, 2023, 15 (04)
  • [2] Few-Shot Hyperspectral Image Classification With Unknown Classes Using Multitask Deep Learning
    Liu, Shengjie
    Shi, Qian
    Zhang, Liangpei
    [J]. IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2021, 59 (06): : 5085 - 5102
  • [3] FEW-SHOT HYPERSPECTRAL IMAGE CLASSIFICATION THROUGH MULTITASK TRANSFER LEARNING
    Qu, Ying
    Baghbaderani, Razieh Kaviani
    Qi, Hairong
    [J]. 2019 10TH WORKSHOP ON HYPERSPECTRAL IMAGING AND SIGNAL PROCESSING - EVOLUTION IN REMOTE SENSING (WHISPERS), 2019,
  • [4] Deep Few-Shot Learning for Hyperspectral Image Classification
    Liu, Bing
    Yu, Xuchu
    Yu, Anzhu
    Zhang, Pengqiang
    Wan, Gang
    Wang, Ruirui
    [J]. IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2019, 57 (04): : 2290 - 2304
  • [5] Deep transformer and few-shot learning for hyperspectral image classification
    Ran, Qiong
    Zhou, Yonghao
    Hong, Danfeng
    Bi, Meiqiao
    Ni, Li
    Li, Xuan
    Ahmad, Muhammad
    [J]. CAAI TRANSACTIONS ON INTELLIGENCE TECHNOLOGY, 2023, 8 (04) : 1323 - 1336
  • [6] A Deep few-shot learning algorithm for hyperspectral image classification
    Liu, Bing
    Zuo, Xibing
    Tan, Xiong
    Yu, Anzhu
    Guo, Wenyue
    [J]. Cehui Xuebao/Acta Geodaetica et Cartographica Sinica, 2020, 49 (10): : 1331 - 1342
  • [7] Heterogeneous Few-Shot Learning for Hyperspectral Image Classification
    Wang, Yan
    Liu, Ming
    Yang, Yuexin
    Li, Zhaokui
    Du, Qian
    Chen, Yushi
    Li, Fei
    Yang, Haibo
    [J]. IEEE GEOSCIENCE AND REMOTE SENSING LETTERS, 2022, 19
  • [8] DEEP SELF-SUPERVISED LEARNING FOR FEW-SHOT HYPERSPECTRAL IMAGE CLASSIFICATION
    Li, Yu
    Zhang, Lei
    Wei, Wei
    Zhang, Yanning
    [J]. IGARSS 2020 - 2020 IEEE INTERNATIONAL GEOSCIENCE AND REMOTE SENSING SYMPOSIUM, 2020, : 501 - 504
  • [9] Deep Cross-Domain Few-Shot Learning for Hyperspectral Image Classification
    Li, Zhaokui
    Liu, Ming
    Chen, Yushi
    Xu, Yimin
    Li, Wei
    Du, Qian
    [J]. IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2022, 60
  • [10] Attention Multisource Fusion-Based Deep Few-Shot Learning for Hyperspectral Image Classification
    Liang, Xuejian
    Zhang, Ye
    Zhang, Junping
    [J]. IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING, 2021, 14 : 8773 - 8788