An improved quantum-behaved particle swarm classifier based on weighted mean best position

Aiming at the weaknesses of PS-classifier, it is easily trapped into locally optimal solution and slow convergence velocity when it deals with the complex problems, an improved quantum-behaved particle swarm classifier has been proposed in the paper. Firstly, It introduce the weighted mean best position to improve the performance of QPSO (quantum-behaved particle swarm), and use a novel Michigan rule to code speech parameters. Then, a new fitness function is constructed to accomplish the weighted quantum-behaved particle swarm classifier (WQPS -classifier). Finally it was applied into speaker recognition. Experimental results show that the proposed classifier achieve higher recognition rate in noisy environments compared with other classification algorithms.