Algorithms for the classification of diseases of paired organs on the basis of neural networks and fuzzy sets
Автор: Zenovich Andrey Vasilyevich, Grebnev Vitaliy Igorevich, Primachenko Filipp Germanovich
Журнал: Математическая физика и компьютерное моделирование @mpcm-jvolsu
Рубрика: Моделирование, информатика и управление
Статья в выпуске: 6 (43), 2017 года.
Бесплатный доступ
In this paper we introduce two algorithms for diagnosing diseases of paired organs by the method of combined radio thermometry. The first one is based on neural networks and the second one is based on the apparatus of fuzzy sets. We consider a new modification of the neural network architecture for constructing the neural network algorithm, which involves the automatic addition of neurons to the output layer during the learning of the neural network. Computational experiments were carried out to diagnose varicose leg diseases and breast diseases. These experiments showed that this modification improves the efficiency of the algorithm by 10-12 %. The diagnostic algorithm based on fuzzy sets on the grounds of diagnosis builds fuzzy sets, after which the diagnosis is set by a method analogous to the method of non-compensatory aggregation. Besides, the algorithm was tested for varicose diseases and breast diseases.
Data mining, microwave radiothermometry, intelligent advisory systems, phlebology, mammalogy
Короткий адрес: https://sciup.org/14969055
IDR: 14969055 | DOI: 10.15688/mpcm.jvolsu.2017.6.3