Comparative study of Landslide Identification using different optimization Algorithms

Abstract

Landslide is a complicated phenomenon related to land movement that cause heavy human loss, ecological imbalance and structural damages. This complicated phenomenon is commonly seen in mountainous regions due to gravitational mass movement and shear strength decrement leading to geological disaster. Here, human activities include excavation, digging, and deforestation; whereas natural calamities includeheavy rainfall, volcanic eruptions and earthquake. Landslide is ranked third among disaster types, as it causes monetary loss to billions of dollars along human loss to millions. Hence, it is compulsory needed to identify landslide to avoid losses in its earlier stage itself. To identify these losses, researchers have established various new methods by utilizing optimization algorithms. The aim of this research is to justify and compare various optimization algorithms for the identification of landslide. The comparison analysis for landslide identification is done with five algorithms, such as Competitive S warm Optimizer (CSO)-basedDeepGenerative Adversarial Network (Deep GAN), Tunicate Swarm Algorithm (TSA)-based deep GAN, Particle Swarm Optimization (PSO) algorithm-based deep GAN, Water Cycle Algorithm (WCA)-based deep GAN, and Water Cycle Particle Swarm Optimization (WCPSO)-based GAN. However, WCPSO is derived by the hybridization of WCA and PSO. From the comparison, the WCPSO exhibits maximum values of accuracy, specificity, and sensitivity with 0.897, 0.857, and 0.915.