Evolutionary algorithm using conditional expectation value for quadratic assignment problem
Wojciech Chmiel · Swarm and Evolutionary Computation · 2019
This paper is devoted to the implementation of the method based on the theoretical properties of the quadratic assignment problem in the implementation of the nature-inspired algorithms. The equation describing the conditional expectation value for the QAP problem objective function was proposed and proved. The conditional expectation was used in the implementation of several types of pseudo-genetic operators. The results obtained with the use of new operators were compared with the results obtained with the use of standard pseudo-genetic operators for the QAP problem. • The two theoretical properties of the QAP problem are proved. • On the basis of the proved properties the twenty-five pseudo-genetic operator are proposed. • All proposed genetic operators for the QAP problem are extensively tested and tuned. • Operators based on the CEV value give a statistically significantly better results than the standard operators.