home bbs files messages ]

Forums before death by AOL, social media and spammers... "We can't have nice things"

   comp.ai      Awaiting the gospel from Sarah Connor      1,954 messages   

[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]

   Message 25 of 1,954   
   Anthony Bucci to All   
   Re: Why are Neural Nets not AI?   
   30 Jul 03 22:07:23   
   
   From: abucci@cs.brandeis.edu   
      
   You should ask your lecturer what he means.  Frankly, it's nigh absurd to   
   say neural networks are not part of AI.   
      
   > This is largely due, IMO, to the work of Minsky & Papert (in their book   
   > Perceptrons). Since that time, their have been numerous developments in   
   > the field - although even then they were only trivial if you limited   
   > yourself to analyzing networks that were easily analyzed (i.e, the   
   > trivial ones).   
      
   What's most ironic and tragic about Perceptrons is that, nowadays, people   
   do a lot of interesting stuff with perceptrons.  There is a lot of image   
   analysis, for instance, begins with a wavelet transform of the image,   
   followed by some kind of component analysis (PCA e.g.), followed by   
   running the resulting coefficients through a perceptron.  Or, SVMs can be   
   thought of as fancy perceptrons -- just add a kernel to a perceptron and   
   POOF! you have an SVM.   
      
   > Ironically, when all is said and done, NNs may end up being the   
   > only enduring part of AI (although I'm sure to get some howls here),   
      
   Oh come on!  This is beyond flippant, especially given the reality that   
   neural nets as a field has virtually imploded.   
      
   About chess.  People fight about this as a test problem.  For all   
   practical purposes, it's not very compelling anymore.  But that's a matter   
   of popular press -- it's not sexy.  However, the AI problem of learning to   
   play chess with minimal human hand holding is far from being solved.   
   Deep Blue does not suffice as a solution by any stretch of the   
   imagination.  I should point out, the AI problem of learning to play   
   TICTACTOE from minimal human input has barely, and in my opinion   
   unconvincingly, been solved.  We have a very long way to go indeed.  As a   
   field, we can hardly agree on how to frame the problems, let alone say   
   what it means to solve them.   
      
   Anthony   
      
   [ comp.ai is moderated.  To submit, just post and be patient, or if ]   
   [ that fails mail your article to , and ]   
   [ ask your news administrator to fix the problems with your system. ]   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)   

[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]


(c) 1994,  bbs@darkrealms.ca