Meysam Orouskhani - Department of Computer Engineering, Science and Research Branch, Islamic Azad University, Tehran, Iran
Mohammad Teshnehlab - Industrial Control Center of Excellence, Electrical Engineering Department, K. N. Toosi University of Technology, Tehran, Iran
Mohammad Ali Nekoui - Industrial Control Center of Excellence, Electrical Engineering Department, K. N. Toosi University of Technology, Tehran, Iran

This paper introduces a novel multi-objective evolutionary algorithm based on cat swarm optimization algorithm (EMCSO) and its application to solve a multi-objective knapsack problem. The multi-objective optimizers try to find the closest solutions to true Pareto front (POF) where it will be achieved by finding the less-crowded non-dominated solutions. The proposed method applies cat swarm optimization (CSO), a swarm-based algorithm with ability of exploration and exploitation, to produce offspring solutions and uses the nondominated sorting method to find the solutions as close as to POF and crowding distance technique to obtain a uniform distribution among the non-dominated solutions. Also, the algorithm is allowed to keep the elites of population in reproduction process and use an oppositionbased learning method for population initialization to enhance the convergence speed. The proposed algorithm is tested on standard test functions (zitzler’ functions: ZDT) and its performance is compared with traditional algorithms and is analyzed based on performance measures of generational distance (GD), inverted GD, spread, and spacing. The simulation results indicate that the proposed method gets the quite satisfactory results in comparison with other optimization algorithms for functions of ZDT1 and ZDT2. Moreover, the proposed algorithm is applied to solve multi-objective knapsack problem.

Multi-objective cat swarm optimization; Non-dominated sorting; Crowding distance; Opposition-based learning