Re: an true information theory

examachine_at_gmail.com
Date: 03/05/05 

Date: 4 Mar 2005 19:12:48 -0800

Daryl McCullough wrote:
> Eray says:
>
> >"Can the rule finding itself be done by some computer program?
> >Interestingly, it is mathematically proven that there can be no
> >computer program which can eventually find (synonym: learn) these
> >(algorithmic) rules for all sequences which have such rules! "
> >
> >Here, the exclamation mark at the end of the first paragraph
suggests
> >that the answer is a NO.
>
> He doesn't just *suggest* that the answer is no, he is stating
> it. However, he is just saying that there is no such thing as
> a *perfect* pattern recognizer program. That isn't saying anything
> about the impossibility of AI, because humans aren't perfect
> pattern recognizers, either.

[snip]

Your theory is kind of interesting. It's a good food for thought.

However, I am not sure that you realize that the author probably refers
to certain no-free-lunch theorems by that sentence with the unnecessary
exclamation mark at the end.

An important point of the algorithmic approach to machine learning is
that the assumptions which cause those no-free-lunch theorems are
removed, e.g. they are not relevant any more. That's why the author
IMHO seems to show lack of understanding of the state-of-the-art in ML.

Regards, 

--
Eray