Artificial immune classifier with swarm learning

Artificial immune systems are computational systems inspired by the principles and processes of the natural immune system. The various applications of artificial immune systems have been used for pattern recognition and classification problems; however, these artificial immune systems have three major problems, which are growing of the memory cell population, eliminating of the useful memory cells in next the steps, and randomly using cloning and mutation operators. In this study, a new artificial immune classifier with swarm learning is proposed to solve these three problems. The proposed algorithm uses the swarm learning to evolve the antibody population. In each step, the antibodies that belong to the same class move to the same way according to their affinities. The size of the memory cell population does not grow during the training stage of the algorithm. Therefore, the method is faster than other artificial immune classifiers. The classifier was tested on two case studies. In the first case study, the algorithm was used to diagnose the faults of induction motors. In the second case study, five benchmark data sets were used to evaluate the performance of the algorithm. The results of second case studies show that the proposed method gives better results than two well-known artificial immune systems for real word data sets. The results were compared to other classification techniques, and the method is competitive to other classifiers.