Dealings with problem hardness in genetic algorithms
Stjepan Picek, Marin Golub · WSEAS Transactions on Computers archive · 2009
Genetic algorithms (GA) have been successfully applied to various problems, both artificial as well as real-world problems. When working with GAs it is important to know those those kinds of situations when they will not find the optimal solution. In other words, to recognize problems that are difficult for a GA to solve. There are various reasons why GAs will not converge to optimal solutions. By combining one or more of these reasons a problem can become a GA-hard problem. Today, there are numerous methods for solving GA-hard problems; every measure has its specific advantages and drawbacks. In this work the effectiveness of one of these measures is evaluated, namely the Negative Slope Coefficient (NSC) measure. A different measure is proposed, called the New Negative Slope Coefficient (NNSC) measure, which aims to address certain drawbacks of the original method. Possible guidelines for further development of this, and comparable methods are proposed.