New Neural Network Algorithm Aimed to Physical Properties of Rocks Prediction Using Log Curves and 3D Seismic Data

Abstract

The article is dedicated to a new neural network algorithm which is aimed to spatial well log curves prediction using seismic data. The specificity of proposed method is usage of variety random functions instead of weight coefficients and activation function in neural networks. As an example of effectivity of the new approach the results of density prediction using two neural network algorithms are demonstrated.