Enhanced imperialist competitive algorithm for 2-satisfiability logic mining in bank marketing data set

Abstract

Imperialist Competitive Algorithm (ICA) is an evolutionary algorithm inspired by the phenomenon of human's socio-political evolution among human empires in the real world, known as imperialistic competition. Meanwhile, logic programming in data mining can explore the underlying relationship in real life data sets. In this paper, enhanced Imperialist Competitive Algorithm is incorporated in the training phase of Hopfield Neural Network to perform the 2-Satisfiability logic mining. Then, the performance of ICA algorithm will be compared with the widely used conventional method, Exhaustive Search (ES) algorithm. The hybrid network will be tested in bank marketing data set. The performance of both algorithms will be evaluated in term of performance evaluation metrics such as Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Sum of Squared Error (SSE), Schwarz Bayesian Criterion (SBC), accuracy and CPU time to determine the effectiveness of the hybrid model. ICA is expected to outperform ES algorithm in doing 2-SAT logic programming.