Predict settlement of gypseous soil under load by deep neural network.

Several techniques have been developed to predict the stability of shallow foundations over the years. However, methods for predicting the necessary degree of detail and consistency have not yet been developed. This study applied the method to a deep neural network (DNN). The expected shallow settlement parameters are carefully selected. It was concluded that a deep neural network (DNN) appears to be a viable solution as it has been used successfully in many diagnostic applications in geotechnical engineering. In this paper, the precipitation values of gypsum soils were predicted under the influence of applied load using the deep learning technique. The study found that this model is very good in predicting soil precipitation and found convergence between real and expected values. The deep neural network model showed a more significant performance with mean absolute and mean squared errors, which were 0.0290 and 0.0387. The DNN model recorded the efficiency and variance computation coefficient at a good rate. It is concluded that a deep neural network model can be used to predict the stability of a shallow foundation on gypseous soil. [ABSTRACT FROM AUTHOR]