New Approach for selecting the most Desirable Options in a Learning Setting