AI-enabled Landslide Recognition for Effective Public Health Risk Management

The frequency of hazardous landslides has increased worldwide as a result of increased heavy rainfall events and increased human construction activities. The assessment of landslide vulnerability is an essential and effective method for preventing landslides. To resolve these issues, this paper develops a novel Rat Swarm integrated Random Forest (RSRF) method to manage the risk evolved through the landslide. The LISS-III satellite dataset of remote sensing (RS) images was gathered for the study. The preprocessing is performed by employing the z-score normalization to standardize the data images. The rat swarm (RS) optimization enhances the feature selection in the landslide by efficiently forecasting relevant assessment and the random forest (RF) employed to improve the classification accuracy in the landslide recognition process. Various existing methods are utilized for the comparison performance with the proposed RSRF techniques. The outcome shows that the proposed RSRF method improved more significantly than all other existing techniques in terms of area under the curve (AUC (93.3%)), F1-score (85.5%), log-loss (40.4%), and recall (97%).