Limiting density of discrete points
||This article needs attention from an expert in Mathematics. (November 2008)|
It was formulated by Edwin Thompson Jaynes to address defects in the initial definition of differential entropy.
Shannon originally wrote down the following formula for the entropy of a continuous distribution, known as differential entropy:
Unlike Shannon's formula for the discrete entropy, however, this is not the result of any derivation (Shannon simply replaced the summation symbol in the discrete version with an integral) and it turns out to lack many of the properties that make the discrete entropy a useful measure of uncertainty. In particular, it is not invariant under a change of variables and can even become negative.
Jaynes (1963, 1968) argued that the formula for the continuous entropy should be derived by taking the limit of increasingly dense discrete distributions. Suppose that we have a set of discrete points , such that in the limit their density approaches a function called the "invariant measure".
Jaynes derived from this the following formula for the continuous entropy, which he argued should be taken as the correct formula:
It is similar to the (negative of the) Kullback–Leibler divergence or relative entropy, which is a comparison between two probability distributions, with one difference. In the Kullback-Leibler divergence, must be a probability density, whereas in Jaynes' formula, is simply a density, meaning that it does not have to integrate to 1.
Jaynes' continuous entropy formula has the property of being invariant under a change of variables, provided that and are transformed in the same way. (This motivates the moniker "invariant measure" for m.) This solves many of the difficulties that come from applying Shannon's continuous entropy formula.
- Jaynes, E. T. (2003). Probability Theory: The Logic of Science. Cambridge University Press. ISBN 978-0521592710.
|This probability-related article is a stub. You can help Wikipedia by expanding it.|