The new negative slope coefficient measure
Stjepan Picek, Marin Golub · 2009
When is a problem easy or difficult for a genetic algorithm? This work focuses on unitation functions as tests for the efficiency of a genetic algorithm in reaching an optimal solution. We research the effectiveness of the Negative Slope Coefficient Measure (NSC measure) in finding difficult problems and present flaws of such a measure. In summary, we present a new measure for defining the hardness of a problem, the new NSC, based on the Fitness Landscape; experimentally we demonstrate the efficacy of the method and compare it with the performance measure achieved by real runs. Finally we propose new steps for development of the method.