Active Algorithms For Preference Learning Problems with Multiple Populations