Identification of Earth’s surface objects using ensembles of convolutional neural networks.

Authors

  • Evgenii E. Marushko United Institute of Informatics Problems, National Academy of Sciences of Belarus, 6 Surhanava Street, Minsk 220012, Belarus
  • Alexander A. Doudkin United Institute of Informatics Problems, National Academy of Sciences of Belarus, 6 Surhanava Street, Minsk 220012, Belarus
  • Xiangtao Zheng Xiʼan Institute of Optics and Precision Mechanics, Chinese Academy of Sciences, Shaanxi, Xiʼan 710119, China

Keywords:

convolutional neural network, support vector machine, neural network ensemble, Earth’s surface image, remote sensing, identification, synthetic aperture radar
Supporting Agencies
This work was partially supported by the Belarusian Republican Foundation for Fundamental Research and the National Foundation of Natural Sciences of China (project No. F20-017).

Abstract

The paper proposes an identification technique of objects on the Earth’s surface images based on combination of machine learning methods. Different variants of multi-layer convolutional neural networks and support vector machines are considered as original models. A hybrid convolutional neural network that combines features extracted by the neural network and experts is proposed. Optimal values of hyperparameters of the models are calculated by grid search methods using k-fold cross-validation. The possibility of improving the accuracy of identification based on the ensembles of these models is shown. Effectiveness of the proposed technique is demonstrated by the example of images obtained by synthetic aperture radar.

Author Biographies

  • Evgenii E. Marushko, United Institute of Informatics Problems, National Academy of Sciences of Belarus, 6 Surhanava Street, Minsk 220012, Belarus

    researcher at the laboratory of identification systems

  • Alexander A. Doudkin, United Institute of Informatics Problems, National Academy of Sciences of Belarus, 6 Surhanava Street, Minsk 220012, Belarus

    doctor of science (engineering), full professor; head of the laboratory of identification systems.

  • Xiangtao Zheng, Xiʼan Institute of Optics and Precision Mechanics, Chinese Academy of Sciences, Shaanxi, Xiʼan 710119, China

    PhD (signal and information processing); associate professor at the key laboratory of spectral imaging technology

References

  1. Kim M, Choi W, Jeon Y, Liu L. A hybrid neural network model for power demand forecasting. Energies. 2019;12(5):931. DOI: 10.3390/en12050931.
  2. Frankel A, Tachida K, Jones R. Prediction of the evolution of the stress field of polycrystals undergoing elastic-plastic deformation with a hybrid neural network model. Machine Learning: Science and Technology. 2020;1(3):035005. DOI: 10.1088/2632-2153/ ab9299.
  3. Liu H, Yang R, Wang T, Zhang L. A hybrid neural network model for short-term wind speed forecasting based on decomposition, multi-learner ensemble, and adaptive multiple error corrections. Renewable Energy. 2021;165:573–594. DOI: 10.1016/j. renene.2020.11.002.
  4. Ma C, Du X, Cao L. Analysis of multi-types of flow features based on hybrid neural network for improving network anomaly detection. IEEE Access. 2019;7:148363–148380. DOI: 10.1109/ACCESS.2019.2946708.
  5. Berkhahn S, Fuchs L, Neuweiler I. An ensemble neural network model for real-time prediction of urban floods. Journal of hydrology. 2019;575:743–754. DOI: 10.1016/j.jhydrol.2019.05.066.
  6. Cheng B, Wu W, Tao D, Mei S, Mao T, Cheng J. Random cropping ensemble neural network for image classification in a robotic arm grasping system. IEEE Transactions on Instrumentation and Measurement. 2020;69(9):6795–6806. DOI: 10.1109/ TIM.2020.2976420.
  7. Large scale visual recognition challenge [Internet; cited 29.01.2021]. Available from: http://image-net.org/challenges/LSVRC/ 2016/results.
  8. LeCun Y, Boser B, Denker JS, Henderson D, Howard RE, Hubbard W, et al. Backpropagation applied to handwritten zip code recognition. Neural computation. 1989;1(4):541–551.
  9. Goodfellow I, Bengio Y, Courville A. Deep Learning. Cambridge: MIT Press; 2016. 781 p.
  10. Parikh D, Polikar R. An ensemble-based incremental learning approach to data fusion. IEEE Transactions on Systems, Man and Cybernetics. Part B: Cybernetics. 2007;37(2):437450. DOI: 10.1109/TSMCB.2006.883873.
  11. Marushko EE, Doudkin AA. Ensembles of neural networks for forecasting of time series of spacecraft telemetry. Optical Memory and Neural Networks. 2017;26(1):47–54. DOI: 10.3103/S1060992X17010064.
  12. Kourentzes N, Barrow D, Crone S. Neural network ensemble operators for time series forecasting. Expert Systems with Applications. 2014;41(9):4235–4244. DOI: 10.1016/j.eswa.2013.12.011.
  13. Vapnik V. The nature of statistical learning theory. 2nd edition. New York: Springer; 1999. 314 p.
  14. Bergstra J, Bengio Y. Random search for hyper-parameter optimization. Machine Learning Research. 2012;13:281305.
  15. Statoil/C-CORE iceberg classifier challenge. Data [Internet; cited 29.01.2021]. Available from: https://www.kaggle.com/c/ statoil-iceberg-classifier-challenge/data.
  16. Kingma DP, Ba J. Adam: a method for stochastic optimization. arXiv:1412.6980. 2017 [cited 29.01.2021]: [15 p.]. Available from: https://arxiv.org/abs/1412.6980.
  17. Simonyan K, Zisserman A. Very deep convolutional networks for large-scale image recognition. arXiv:1409.1556 [Preprint]. 2015 [cited 29.01.2021]: [14 p.]. Available from: https://arxiv.org/abs/1409.1556.
  18. Chollet F. Xception: deep learning with depthwise separable convolutions. In: IEEE Computer Society. 2017 IEEE Conference on computer vision and pattern recognition (CVPR); 2017 July 21–26; Honolulu, USA. Los Alamitos: IEEE; 2017. p. 1251–1258. DOI: 10.1109/CVPR.2017.195.
  19. He K, Zhang X, Ren S, Sun J. Deep residual learning for image recognition. In: IEEE Computer Society. Proceedings of the IEEE conference on computer vision and pattern recognition; 2016 June 27–30; Las Vegas, Nevada. Los Alamitos: IEEE; 2016. p. 770–778. DOI: 10.1109/CVPR.2016.90.
  20. Tan M, Le QV. Efficient net: rethinking model scaling for convolutional neural networks. arXiv:1905.11946 [Preprint]. 2020 [cited 29.01.2021]: [11 p.]. Available from: https://arxiv.org/abs/1905.11946.
  21. Huang G, Liu Z, Van Der Maaten L, Weinberger KQ. Densely connected convolutional networks. In: IEEE Computer Society. 2017 IEEE Conference on computer vision and pattern recognition (CVPR); 2017 July 21–26; Honolulu, USA. Los Alamitos: IEEE; 2017. p. 2261–2269. DOI: 10.1109/CVPR.2017.243.
  22. Sandler M, Howard A, Zhu M, Zhmoginov A, Chen LC. MobilenetV2: inverted residuals and linear bottlenecks. In: IEEE Computer Society. 2018 IEEE/CVF Conference on computer vision and pattern recognition; 2018 June 18–23; Salt Lake City, USA. Los Alamitos: IEEE; 2018. p. 4510–4520. DOI: 10.1109/CVPR.2018.00474.
  23. Prechelt L. Early stopping – but when? In: Orr GB, Müller K-R, editors. Neural Networks: tricks of the trade. Berlin: Springer; 1998. p. 55–69.
  24. Goyal P, Dollar P, Girshick R, Noordhuis P, Wesolowski L, Kyrola A, et al. Accurate, large minibatch SGD: training imagenet in 1 hour. arXiv:1706.02677 [Preprint]. 2018 [cited 29.01.2021]: [12 p.]. Available from: https://arxiv.org/abs/1706.02677.

Downloads

Additional Files

Published

2021-08-05

Issue

No. 2 (2021): Journal of the Belarusian State University. Mathematics and Informatics

Section

Informatics, Computer Science and Management

License

The authors who are published in this journal agree to the following:

  1. The authors retain copyright on the work and provide the journal with the right of first publication of the work on condition of license Creative Commons Attribution-NonCommercial. 4.0 International (CC BY-NC 4.0).
  2. The authors retain the right to enter into certain contractual agreements relating to the non-exclusive distribution of the published version of the work (e.g. post it on the institutional repository, publication in the book), with the reference to its original publication in this journal.
  3. The authors have the right to post their work on the Internet (e.g. on the institutional store or personal website) prior to and during the review process, conducted by the journal, as this may lead to a productive discussion and a large number of references to this work. (See The Effect of Open Access.)

How to Cite

[1]
Marushko, E.E. et al. 2021. Identification of Earth’s surface objects using ensembles of convolutional neural networks. Journal of the Belarusian State University. Mathematics and Informatics. 2 (Aug. 2021), 114–123. DOI:https://doi.org/10.33581/2520-6508-2021-2-114-123.