Message-ID: <80a89bc00809261138u300786d7h3502a91ddf7d8bbb@mail.gmail.com>
Date: 2008-09-26T18:38:12Z
From: Vishal Belsare
Subject: Total Correlation / Multi-Information
Hi,
Given a set of stochastic sequences, say n with T observations on
each. Is there any algorithm (hopefully implemented) to compute the
total correlation / multi-information for this set of random
sequences? I know that it entails estimation of entropy, and that is
where I see blurry. How does one bin the data? Or does one estimate a
histogram?
Apart from R pointers, it would be great if anyone can also point to
relevant papers / references on this.
I am looking to work with this in R and am hoping that there are many
others who have done this before. I have taken a look at CRAN, and
came across the entropy and minet package; intend to see how I could
utilize those. Thanks much in anticipation.
Best Regards,
Vishal Belsare
--
Firstly, I propose for myself the rule that the obviousness of a
proposition does not free me from the obligation to continue searching
for a proof of it, at least until I clearly realize that absolutely no
proof could ever be required, and why.
~ Bolzano