learning

Shannon entropy estimation in countably infinite alphabets is addressed by adopting convergence results of the entropy functional and concentration inequalities.

A new histogram-based mutual information estimator using data-driven tree-structured partitions (TSP) is presented in this work.