Can anyone suggest the best way of using an asymmetric similarity measure (like, for example, general NCM) for kNN?

Similar questions and discussions