A directed quantile cross-entropy method for 0/1 knapsack problems
Libao Deng, Liyan Qiao, Xiyuan Peng · 2009
Cross-entropy method is base on probability density function. It is robust, easy to use. With analysis of advantages and disadvantages of the cross-entropy method, a directed quantile method based on cross-entropy is proposed. The main idea of the directed quantile cross-entropy method is to select alterable quantity vectors using for producing a ldquobetterrdquo sample in the next iteration. The convergence speed and search best result of the directed quantile cross-entropy are tested using 0/1 knapsack problems. The experiments show that the search efficiency of the modified cross-entropy method is more significantly improved than quantum-inspired evolutionary algorithm and cross-entropy method.