Tackling 0/1 knapsack problem with gene induction

Abstract

We propose a gene induction approach for genetic algorithms. It is more robust compared to the traditional approach in genetic algorithms. The approach was applied to 0/1 knapsack problem. It found near optimal results in all the representative problem instances reported in the literature, while traditional approaches failed in a number of instances because of preponderance of infeasible individuals in the population. In combination with a heuristic mutation operator, our method provided better results for all the problem instances investigated.