Acquisition of pattern classification rule based on particle swarm optimization

This paper presented a rule extraction algorithm based on particle swarm optimization. A single rule was encoded as a particle; through its velocity-position search model and the information stored by the particles, the optimal classification rule set was generated. The proposed algorithm was used to generate pattern classification rules for Iris data set. The simulation results showed that the classification rules generated by the proposed algorithm achieved more classification accuracy and lower computational cost than the other methods.