Abstract:Purpose. To compare effeciency of different realizations of approaches to estimation of mutual information function based on nearest neighbours.
Method. Two approaches to calculation of mutual information function were realized numerically: straightforward approach is based on brute force, and sorting based one.
Results. The algorithmic complexity of sorting beased method was shown to be less than of straightforward approach, but larger than the complexity of any quick sort method.
Discussion. Realization of sorting based method is reasonable in the case, when one has to deal with long samplings, while for small samplings the straightforward approach is enough.