We present formal definitions of two commonly observed asymmetries in a concave receiver operating characteristic curve. The main theorem of the paper proves that the Kullback– Leibler divergences between the underlying signal and noise variables are ordered based on these asymmetries. This result is true for any continuous distributions of the signal and noise variables. © 2015 Elsevier B.V. All rights reserved.
- Kullback-Leibler divergence
- Relative distributions