A highly optimized multi-stage teacher-learner inspired particle swarm optimizer system

The method of Particle Swarm Optimization (PSO) is an epitome in optimization process for solving classification, clustering, and many other multi-dimensional problems. PSO is found to be stronger than few state of the art population-based coup such as Ant colony optimization (ACO), genetic algorithms (GAs), Optimization of the herd of elephants known as Elephant herd optimization (EHO) together with other algorithms under certain real-life scenarios like path finding, feature selection, etc. Originally PSOs were implemented by means of optimizing the velocity/speed and position update equations so that it could locate the best global position of the particles. Many researchers have designed different approaches by combining human behavioral aspects to these equations in order to further optimize the PSOs performance. In this text we have proposed the use of a 4-staged teacher-learning behavior (TLB) inspired PSO for optimizing the system performance. Under different test functions, the proposed algorithm was tested and evaluated and its performance was compared relative with the modern state of the art PSO systems. It is observed that the proposed PSO reduces the convergence time by more than 20%, and also reduces the overall computational complexity to half. We conclude this text by making some acute observations about the implemented system, and recommend methods which can be used for further optimizing the system performance.