Non-logarithmic Jensen-Shannon divergence

A.P. MAJTEY; P.W. LAMBERTI

Colonia

Congreso; XIII Meeting on Nonequilibrium Statistical Mechanics and Nonlinear Physics; 2002

J. Lin introduced an information-theoretic based divergence measure between two or more probability distributions. This measure is known as de Jensen-Shannon divergence (JSD). Its main property is that it does not require the condition of absolute continuity for the probability distributions involved. In this work we investigate the properties of a generalized version of the JSD in the framework of Tsallis statistics. This kind of generalization has been investigated for other divergence measure.