Output details

A multi-objective GRASP for partial classification

<15> This is a new application of the greedy randomized adaptive search procedure (GRASP) to Multi-Objective Optimisation. Unlike previous applications, it is not based on repeated optimisation of a single objective. Instead, our approach is guided by Pareto-optimality and dominance, which extracts improved classification rules. Others have adapted our approach, e.g. Arroyo and Pereira (Int. J. Advanced Manufacturing Technology 2011, 55(5-8), 741-753) also use dominance; Duarte et al. (Computers & Operations Research, 2010, 37(3), 498-508) provide a version using path relinking; Salazar-Aguilar et al. (J. Heuristics, 2013, 19, 179—200) develop a dominance based GRASP for a bi-objective design problem.