A novel bacterial algorithm to extract the rule base from a training set

In this paper a novel bacterial algorithm to extract the rule base starting from a training set is presented. The proposed algorithm also optimizes the input and output membership function parameters. The algorithm is based on the use of bacterial operations on every rule in the rule set. A reduced optimized rule base is obtained by using rule fusion and removal procedures. The algorithm performance was evaluated by using a six input variables target function frequently used in the literature as benchmark. The obtained results show good performance with respect to the works recently presented in the literature