Newsgroups: comp.ai
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!fas-news.harvard.edu!newspump.wustl.edu!news.ecn.bgu.edu!vixen.cso.uiuc.edu!howland.reston.ans.net!news.sprintlink.net!simtel!harbinger.cc.monash.edu.au!bunyip.cc.uq.oz.au!marlin.jcu.edu.au!mccane
From: mccane@cs.jcu.edu.au (Brendan McCane)
Subject: Re: Machine Learning Breakthrough
Message-ID: <mccane.801449277@reef.cs.jcu.edu.au>
Sender: news@marlin.jcu.edu.au (USENET News System)
Organization: James Cook University
X-Newsreader: NN version 6.5.0 #4
References: <3pqr22$5s8@usenet.srv.cis.pitt.edu> <3pqvuj$ddb@newsbf02.news.aol.com>
Date: 26 May 95 00:47:57 GMT
Lines: 76

In <3pqvuj$ddb@newsbf02.news.aol.com> marengom@aol.com (MarengoM) writes:

>Re: Machine Learning Breakthrough

>On 22 May 95 01:21:32 GMT mccane@cs.jcu.edu.au (Brendan McCane) writes:
>Message-ID: <mccane.801105692@reef.cs.jcu.edu.au>

>> [...]
>>Here are my thoughts on the current thread and the AXIS network. Firstly,
>>I have no problems with the AXIS concepts per se and the idea of having
>>a system that can modify its own learning process is an interesting one. 
>>What I do have problems with is the lack of any real information in any 
>>of your posts.  OK, fine you've explained what the network is and that 
>>the learner does the learning and another part does conflict resolution 
>>and blah blah blah. But you haven't given any concrete information.

>  Now this is simply not true.  Judging by some of our email, we have
>given more than enough concrete information about AXIS for anyone to begin
>writing their own programs and conduct their own experiments.  If
>everything has not been disclosed, the information is there to permit the
>rest to be extrapolated.  Allow me to list briefly some the items we have
>disclosed:

/* some general info deleted */

I still think you've missed my point. Yes you have provided information,
but mostly it has been in the form of concepts of the AXIS system. From
the thread I have an understanding of how the thing works in a general
sort of way. Nevertheless, no algorithms have been provided and that
makes it very difficult for me to reproduce any of your work. And if I
can't reproduce any of your work, why should I believe that you have
indeed made a "Machine Learning Breakthrough". I agree that I could
probably go ahead and design my own AXIS system, but it's not going to
be the same as yours - and if the results from it are not good, I would
have wasted my time.

>  It is inconsiderate to maintain that we have not provided enough "real
>information", unless you have not had the opportunity to follow the thread
>from the start.  We believe we have provided more than enough!  And we are
>on record as having promised that we will continue to provide more.

My point is that you can't claim to have made a "Machine Learning
Breakthrough" on the ideas of AXIS alone. You need to provide both
algorithms and RESULTS.

>> [...] Until you address this lack of information by providing
>>algorithms and results, there are going to be a lot of people who won't
>>take you seriously and you can hardly make the claim of having come up
>>with a revolutionary "Machine Learning Breakthrough" and expect people
>>to believe you!

>  From our point of view, it really does not matter that "there are going
>to be a lot of people who won't take [us] seriously".  What matters is
>that *a few* will and more than a few already have.  What you really want
>is to have a program delivered to you on a platter.  So be it.  But be
>patient.  It is our program and we reserve the right to publish only when
>we feel it's ready. :-)

No I don't want a program delivered to me on a platter. But I would like
to see some algorithms and some results. If you don't have any results,
how do you know that your idea is in fact a breakthrough (how do I know
for that matter). Ideas are all well and good, but many seemingly good
ideas just don't work particularly well in practical situations.

>Louis Savain

>Marengo Media, Inc.

Cheers,

Brendan.
--
----------------------------------------------------------------------------
Brendan McCane                            Email:  mccane@coral.cs.jcu.edu.au
C.S. Dept., James Cook University,        Phone:  (077) 815849.
Townsville, QLD, 4811.  Australia.        There's only one catch - Catch 22.
