Open AccessIdentification of Earth’s surface objects using ensembles of convolutional neural networks.
Author(s) -
Е. Е. Марушко,
Alexander Doudkin,
Xiangtao Zheng
Publication year - 2021
Publication title -
žurnal belorusskogo gosudarstvennogo universiteta. matematika, informatika/žurnal belorusskogo gosudarstvennogo universiteta. matematika, informatika
Language(s) - English
Resource type - Journals
eISSN - 2617-3956
pISSN - 2520-6508
DOI - 10.33581/2520-6508-2021-2-114-123
Subject(s) - hyperparameter , convolutional neural network , computer science , hyperparameter optimization , artificial intelligence , identification (biology) , artificial neural network , pattern recognition (psychology) , support vector machine , machine learning , botany , biology
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.