An Extensive Study on Gravitational Search Algorithm

Abstract

Gravitational search algorithm is a naturally occurring algorithm based on Newton's mathematical model of the law of gravitation and motion. Over the course of a decade, researchers have provided many variants of the gravitational search algorithm by modifying its parameters to effectively solve complex optimization problems. This paper conducts a comparative analysis of ten types of gravity search algorithms that modify the three parameters of optimum, speed and position. Tests are conducted on two sets of benchmark types, namely standard functions and issues belonging to different types such as CEC2015 functions, univocal, multimodal and unrestricted optimization functions. Performance comparison is evaluated and statistically validated based on the average exercise value and concentration graph. In trials, IGSA has achieved excellent accuracy through a balanced trade between exploration and exploitation. Furthermore, three negative breast cancer datasets were considered to analyze the efficacy of GSA variants for the black section. Different performance analyzes were performed based on both quality and quantity with the integrated jacquard index as a performance measure. Tests confirm that the IGSA based method worked better than other methods.