A New Estimator for Shannon Entropy

  • Havva Alizadeh Noughabi
  • Jalil Jarrahiferiz
Keywords: Information theory, sample entropy, spacings, local linear model, mean squared error

Abstract

In this paper, we propose a new estimator for Shannon entropy of a continuous random variable. Consistency and other properties of the new estimator is stated. Through a Monte Carlo simulation, the mean squared error of the proposed estimator is compared with some prominent estimators, namely Vasicek’s estimator, Van Es’s estimator, and Correa’s estimator. Finally, it is shown that the proposed estimator has smaller mean squared error than the other estimators.

Published
2019-07-10
Section
Articles