Psychological model of particle swarm optimization based multiple emotions

This paper proposes a novel approach to swarm particle optimization based on emotional behavior to solve real optimization problems. In the trend of PSO manipulating self-adaptive control to regulate potential parameters, the proposed algorithm involves both a semi-adaptive inertia weight and an emotional factor at the level of the velocity rule. The semi-inertia weight highlights a specific comportment. Thus, due to the few changes occurred in its adaptive "life", it continues to evolve with a significantly smaller constant for the benefit of a finer exploitation. The emotion factor presents an important feature of convergence because it splits up the search space into potential regions that are finely explored by sub-swarm populations with the same emotions. The principle of particles with multiple emotions intended for the categorization of particles into specific emotional classes. The idea behind this principle is to divide to conquer, and due to presence of multiple emotional classes the multidimensional search space is widely explored at the search of the best position. Emotional PSO is evaluated on the test suit of 25 functions designed for the special session on real optimization of CEC 2005, and its performances are compared to the best algorithm the restart CMA-ES.