Solving the 0/1 Knapsack Problem Using a Galactic Swarm Optimization with Data-Driven Binarization Approaches

Abstract

Metaheuristics are used to solve high complexity problems, where resolution by exact methods is not a viable option since the resolution time when using these exact methods is not acceptable. Most metaheuristics are defined to solve problems of continuous optimization, which forces these algorithms to adapt its work in the discrete domain using discretization techniques to solve complex problems. This paper proposes data-driven binarization approaches based on clustering techniques. We solve different instances of Knapsack Problems with Galactic Swarm Optimization algorithm using this machine learning techniques.