An improved TLBO leveraging group and experience learning concepts for global functions

Abstract

In this paper, we have proposed a variant of teaching-learning-based optimization (TLBO) algorithm, which leverages group and experience learning of different learners to enhance the overall performance of the original TLBO algorithm. The modified algorithm is based upon the concept of microteaching in which a class (called population) is divided into different smaller groups (called sub-populations) and algorithm is individually run for all the sub-populations before being merged together after certain generations to improve the diversity of the population. Within each sub-population, the algorithm uses the mean values of all the learners within that group, and exploiting other individuals learning experience to find the optimum value. The algorithm hugely benefits of exploitation perspective by strategizing group concept incorporated in the algorithm. Whereas the exploration search immensely benefits from randomly regrouping of sub-populations and learning experience mechanisms inculcated into the algorithm. The proposed algorithm is tested on several bench mark function which proved that GTLBOLE has some good performances when compared with other established algorithms including other variants of teaching-learning-based optimization techniques.