Re: Entropy and Quantity of Information
From: David Cressey <cressey73_at_verizon.net>
Date: Sat, 12 Jan 2008 09:17:40 GMT
Message-ID: <UA%hj.248$ac7.143_at_trndny03>
Date: Sat, 12 Jan 2008 09:17:40 GMT
Message-ID: <UA%hj.248$ac7.143_at_trndny03>
"Joe Thurbon" <usenet_at_thurbon.com> wrote in message news:u_Thj.2146$421.324_at_news-server.bigpond.net.au... [snip good stuff]
> I vaguely recall that some of the more theoretical machine learning
> results (like learnability and optimality results) rely on a notion
entropy.
I once wrote a little program to play Mastermind (a code guessing game) using entropy to measure a move's worth. Mastermind can be played using a brute force tree search. But my program tried out only about 10 moves, and picked the one with the best entropy. Its play was only slightly inferior to the play of a brute force tree searcher. Received on Sat Jan 12 2008 - 10:17:40 CET