Guided imitation optimizer: a metaheuristic combining guided search and imitation search

Abstract

This paper proposes a novel metaphor-free metaheuristic, namely the guided imitation optimizer (GIO). This metaheuristic combines the guided search and imitation-based search. There are five guided searches and three imitation based searches. Meanwhile, there are three references used in this metaheuristic: global finest, a randomly picked solution among the swarm, and a randomized solution within the search space. GIO is then evaluated by using 23 classic functions that consist of seven high dimension unimodal functions (HDUF), six high dimension multimodal functions (HDMF), and ten fixed dimension multimodal functions (FDMF). Through simulation, GIO is superior to golden search optimizer (GSO), grey wolf optimizer (GWO), puzzle optimization algorithm (POA), and coati optimization algorithm (COA) in handling most of these functions. GIO is the first finest in tackling seventeen functions and second finest in tackling six functions. Tight competition occurs between GIO and COA due to the performance of COA which becomes the second finest in handling most of these functions.