Landslide Susceptibility Assessment by Using Convolutional Neural Network

This study performs a GIS-based landslide susceptibility assessment using convolutional neural network, CNN, in area of the Gorzineh-khil region, northeastern Iran. For this assessment, 15-layered CNN was programmed Python high-level language for mapping. In regard, as far landside triggering factors are concerned, it concluded that geomorphologic/topographic parameters (i.e., slope curvature, topographical elevation, aspect, and weathering) water condition (hydrological gradient, drainage pattern, flow gradient) main factors. These provided dataset, which input to CNN. We used 80% dataset training remaining 20% testing prepare map area. order cross-validate resulting map, loss function, common classifiers were considered: support vector machines, SVM, k-nearest neighbor, k-NN, decision tree, DT. An evaluation results revealed led other classes terms 79.0% accuracy, 73.0% precision, 75.0% recall, 77.0% f1-score, and, hence, better accuracy least computational error when compared models.