Abstract
The use of a neural network approach is associated with high requirements for software implementation. The main criteria is a high degree of accuracy in the recovery of the desired geological-physical model of the environment (GPME) and processing speed. The paper presents a description of the developed neural network architecture for solving the inverse problem of geophysics for determining the geometric properties of GPME objects. The performance of a single neural network and an ensemble of neural networks (NN) has been evaluated. The results are presented comparing the operating time of the NN when restoring models on various computing devices: CPU and GPU. The results of experiments on the restoration of various GPME using the developed NN based on the LSTM layer and the U-Net architecture are presented. This work was supported by the RFBR grant No. 19-07-00170.
Export citation and abstract BibTeX RIS
Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.