Class-dependent, non-convex losses to optimize precision.

David M. J. Tax, Feng Wang

Browse the full ICPR paper archive.