Clonal selection based parameter optimization for sparse fuzzy systems

Nature inspired algorithms proved to be very advantageous in several application areas. This paper presents the application of the clonal selection algorithm (originated from the functioning of the vertebrate's immune system) for the tuning of fuzzy inference systems. The proposed solution was tested in case of SISO and MISO systems with two different types of initialization. In each case the performance of the fuzzy system was improved significantly at the end of the tuning process. The resulting parameter sets were validated against test data sets.