Rule induction based on a novel evolutionary strategy

With the development of data mining research, we gradually realize that the main task of the machine learning algorithm is not to find the knowledge without any error, but to find some novel and easy to understand knowledge, which is used for decision making in a higher level through people´s participation. Based on this consideration, a novel data-mining method, an evolutionary strategy with immune selection and lifecycle, is designed for high level rule induction. In this novel algorithm, an immune selection operator and the crossover and mutation operators with lifecycle are employed, in which the former is used for resisting the degeneracy of the original evolutionary strategy, and the latter is for adjusting the probabilities of the crossover and mutation operators more effectively. Simulations on a large database show that the above method has good validity and rationality on rule induction.