carpathian_2004_20_051_066_001

On an information divergence measure and information inequalities


Pranesh KumarLaura Hunter


Abstract

carpathian_2004_20_051_066_abstract

Full PDF

carpathian_2004_20_051_066

Information divergence measures are often useful for comparing two probability distributions. An information divergence measure is suggested. This measure is symmetric in respect of the probability distributions and belongs to the well known class of Csisz´ar’s f-divergences. Its properties are studied and bounds in terms of often used information divergences are obtained. A numerical illustration based on symmetric and asymmetric probability distributions compares some of these divergence measures.

Additional Information

Author(s)

Hunter, Laura, Kumar, Pranesh