Mark: > I realize one can use use a training map to cluster like features, > but is there a way to have a "leftover" class that throws everything > else that doesnt match a defined class into this "leftover" category? ie you want an "outliers" class? Hamish