Distribution of Mutual Information from Complete and Incomplete Data
Keywords: Mutual information, cross entropy, Dirichlet distribution, second
order distribution, expectation and variance of mutual
information, feature selection, filters, naive Bayes classifier,
Bayesian statistics.
Abstract: Mutual information is widely used, in a descriptive way, to measure the
stochastic dependence of categorical random variables. In order to address
questions such as the reliability of the descriptive value, one must consider
sample-to-population inferential approaches. This paper deals with the
posterior distribution of mutual information, as obtained in a Bayesian
framework by a second-order Dirichlet prior distribution. The exact analytical
expression for the mean, and analytical approximations for the variance,
skewness and kurtosis are derived. These approximations have a guaranteed
accuracy level of the order O(1/n^3), where n is the sample size. Leading order
approximations for the mean and the variance are derived in the case of
incomplete samples. The derived analytical expressions allow the distribution
of mutual information to be approximated reliably and quickly. In fact, the
derived expressions can be computed with the same order of complexity needed
for descriptive mutual information. This makes the distribution of mutual
information become a concrete alternative to descriptive mutual information in
many applications which would benefit from moving to the inductive side. Some
of these prospective applications are discussed, and one of them, namely
feature selection, is shown to perform significantly better when inductive
mutual information is used.
BibTeX Entry
@Article{Hutter:04mifs,
author = "Marcus Hutter and Marco Zaffalon",
title = "Distribution of Mutual Information from Complete and Incomplete Data",
_number = "IDSIA-11-02",
journal = "Computational Statistics \& Data Analysis",
volume = "48",
number = "3",
pages = "633--657",
year = "2005",
_month = mar,
publisher = "Elsevier Science",
url = "http://www.hutter1.net/ai/mifs.htm",
url2 = "http://arxiv.org/abs/cs.LG/0403025",
ftp = "ftp://ftp.idsia.ch/pub/techrep/IDSIA-11-02.pdf",
categories = "I.2. [Artificial Intelligence]",
keywords = "Mutual information, cross entropy, Dirichlet distribution, second
order distribution, expectation and variance of mutual
information, feature selection, filters, naive Bayes classifier,
Bayesian statistics.",
abstract = "Mutual information is widely used, in a descriptive way, to measure the
stochastic dependence of categorical random variables. In order to address
questions such as the reliability of the descriptive value, one must consider
sample-to-population inferential approaches. This paper deals with the
posterior distribution of mutual information, as obtained in a Bayesian
framework by a second-order Dirichlet prior distribution. The exact analytical
expression for the mean, and analytical approximations for the variance,
skewness and kurtosis are derived. These approximations have a guaranteed
accuracy level of the order O(1/n^3), where n is the sample size. Leading order
approximations for the mean and the variance are derived in the case of
incomplete samples. The derived analytical expressions allow the distribution
of mutual information to be approximated reliably and quickly. In fact, the
derived expressions can be computed with the same order of complexity needed
for descriptive mutual information. This makes the distribution of mutual
information become a concrete alternative to descriptive mutual information in
many applications which would benefit from moving to the inductive side. Some
of these prospective applications are discussed, and one of them, namely
feature selection, is shown to perform significantly better when inductive
mutual information is used.",
note = "to appear",
}