Design and development of binary metaheuristic algorithms for Feature Selection

Abstract

There are many real life complex problems existing in this world which need to be taken into consideration. To solve such kind of problems various techniques have been proposed. There are many classical techniques designed to solve them. But there are some drawbacks associated with these techniques. Therefore in order to solve the complex problems efficiently, metaheuristic techniques have been designed and developed. Spotted hyena optimizer (SHO) and Emperor penguin optimizer (EPO) are the metaheuristic algorithms developed to solve the optimization problem. The main motive of the thesis work is to design and develop the binary version of spotted hyena optimizer (BSHO) and Emperor penguin optimizer (BEPO). Spotted Hyena Optimizer (SHO) is a recently developed metaheuristic technique that mimics the hunting behavior of the spotted hyenas. The three main steps of this algorithm are prey searching, encircling, and attacking. In the BSHO algorithm, tangent hyperbolic function is utilized to squash the continuous position and then these values are used to update the position of spotted hyenas. Emperor Penguin Optimizer is a novel meta-heuristic algorithm which is inspired from the huddling behavior of the emperor penguins. In the binary version of this algorithm, we have used the sigmoidal transfer function to update the positions of the emperor penguins. Both these algorithms have been tested on 29 benchmark test functions and compared with the other binary metaheuristic algorithms. They have been applied on the feature selection problem also.