Tezislər / Theses

 
Ədəbiyyat 
1. Kadirov F.A. Application of the Hartley transform for Interpretation of Gravity Anomalies in 
the Shamaky – Gobustan and Absheron Oil and Gas Bearing Regions, Azerbaijan. Journal 
of Applied Geophysics, v.45, 2000, p.49 – 61. 
2. Rao B.N., Rama K.P., Markandeyulu A., Mapros – A computer program for basement 
mapping and filtering of gravity and magnetic data using a Hartley transform. Computer 
and Geosiences , 1995. V. 22, №3, p.197 – 218. 

---

THE 3
rd
 INTERNATIONAL SCIENTIFIC CONFERENCES OF STUDENTS AND YOUNG RESEARCHERS 
dedicated to the 99
th
anniversary of the National Leader of Azerbaijan Heydar Aliyev
92

---

 
 
 
 

---

THE 3
rd
 INTERNATIONAL SCIENTIFIC CONFERENCES OF STUDENTS AND YOUNG RESEARCHERS 
dedicated to the 99
th
anniversary of the National Leader of Azerbaijan Heydar Aliyev
94
APPLICATION OF A NEURAL NETWORK IN 
ELECTROGRASTROGRAPHY 
Gunel Agayeva 
Azerbaijan Technical University 
Baku, Azerbaijan 
[e-mail] 
Supervisor: Associate Professor Namik Abdullaev 
Keywords: 
neural
network, electrogastrography, classification of diseases, gastric 
ulcer. modular structure. 
Informative signs obtained to study the functional status of the organs of 
the gastrointestinal tract and to classify diseases can be used as input 
parameters of the artificial neural network. [1] 
Topological analysis of artificial neural networks, analysis of structures, 
analysis of activation functions, teaching methods and algorithms, as well as 
the study of methods for normalization of input parameters justify the choice 
of a multilayer perceptron scheme. Consistent study of the characteristics of 
the input images through a multilayer perceptron scheme and more efficient 
classification are noted as positive features of this structure. The analysis of 
various algorithms confirms the advantage of the algorithm "Back 
propagation" during the operation of the learning phase of the artificial neural 
network. The neural network learning technology with the teacher is more 
suitable for the classification of input images. The structure of the multilayer 
perceptron for the study of normal and ulcerative diseases of the 
gastrointestinal tract One of the most important problems in such structures 
is to determine the number of artificial neurons in the intermediate layers, 
because the metrological parameters of the neural network (accuracy and 
generalized capacity of the system) depend on this value. In 
electrophysiological studies, a structure of a modular neural network based 
on a multilayered perceptron has been proposed to classify the functional 
status of organs.
The structure of the proposed modular-type artificial neural network to 
study the functional status of the gastrointestinal tract is presented in Figure 
1. Fourier complex coefficients (Gi) are used as input parameters. It should 
be noted that the effective (Pi) value of the electrogastroenterographic signal 
can also be used as an additional input parameter to minimize the error of 
the diagnostic operation [2 ]. The main advantage of the modular structure is 
that each module is focused on the diagnosis of a disease, which reduces 
the likelihood of misdiagnosis and increases the accuracy of the whole 
system. On the other hand, the introduction of a modular structure increases 
the functional capacity of the artificial neural network as a result of the 

---