A new optimization algorithm inspired by the quest for the evolution of human society: Human felicity algorithm

Abstract

The present paper introduces a new evolutionary algorithm to solve optimization issues called the human felicity algorithm (HFA). The main idea of the HFA has been inspired the efforts of human society to become felicity. Because the change in human felicity is possible, only through a change in human thought, in the proposed algorithm, the objective function is human felicity in society and the search space optimization is the human thought in society. Like other optimization algorithms, the human felicity algorithm starts solving the problem by an initial random population. Then the population is divided into three categories of elites, disciples, and ordinary people. Afterward, people in human society change their thoughts by 3 ways: 1- being influenced by elites 2- personal experiences 3- sudden and drastic changes of thoughts for various reasons, such as war and famine. To test the capability of the proposed algorithm, the standard test functions of IEEE (Institute of Electrical and Electronics Engineers) Congress on Evolutionary Computation (CEC) 2014, CEC 2019, and CEC2020 with different dimensions have been used. In addition, to evaluate the ability of the proposed algorithm to solve various real-world complex engineering problems, 4 standard classic engineering optimization problems are examined and compared. To test the ability of the proposed algorithm in hyper-parameters tuning, the five-database classification was simulated using the Support Vector Machines (SVM). The simulation of the proposed algorithm on different test functions shows its high ability to solve complex problems with many local optimal points. For statistical analysis, powerful nonparametric methods such as Friedman and Nemenyi post-Hoc test for statistical analysis have been used. Comparing the HFA with some other optimization algorithms demonstrates its better performance in solving complicated multidimensional problems with the lower iteration.