Social interaction optimised Swarm Intelligence technique for Travelling Salesman Problem

The collective intelligent behavior demonstrated by a swamp of simple natural agents often called Swarm Intelligence (SI) is already an emerging area of research. SI has been successfully adapted and applied to numerous problems in the field of combinatorial optimisation. Yet there is space for improvement in these heuristics and in the resulting algorithms as existing SI models still suffer from deficiencies in solution accuracy, speed of convergence and premature convergence to non-optimal solutions. Researchers attempt to hybridize such basic algorithms in order to overcome the weaknesses when the algorithms are applied individually. Among them, Particle Swarm Optimisation and Ant Colony Optimisation are often used by researchers due to their simplicity, effectiveness and efficiency in problem solving. This paper presents a socially optimised hybrid version of Particle Swarm Optimisation and Ant Colony Optimisation that adapts social behaviours existing in swarm population. Experimental results show that the objective of reducing the delay in convergence while maintaining an acceptable solution quality by incorporating social interactions in to Swarm Intelligence models is successful when solving Travelling Salesman Problems.