Genetic algorithm for optimal multivariate mixture

This paper proposes an algorithm to find an optimal mixture that is as close as possible to an ideal solution, starting from a set of elements (items) described by a set of variables (features). This class of optimization problems can be tackled through traditional approaches belonging to the field of operations research (OR) or even through meta-heuristics techniques belonging to the field of artificial intelligence (AI). In order to present an artificial intelligence perspective, this paper uses a genetic algorithm (GA) model which proves its consistency through the comparison with a linear programming (LP) solver on a set of 8-items 5-features experiments. Results show that the proposed GA converges towards the global optimum and provides competitive results