Re: Hawkins ideas on building AI's
From: bkaz (bkaz__at_hotmail.com)
Date: 10/25/04
- Next message: David Longley: "Re: Three Dog Problems"
- Previous message: Glen M. Sizemore: "Re: Hawkins ideas on building AI's"
- In reply to: dan michaels: "Re: Hawkins ideas on building AI's"
- Next in thread: Eray Ozkural exa: "Re: Hawkins ideas on building AI's"
- Reply: Eray Ozkural exa: "Re: Hawkins ideas on building AI's"
- Messages sorted by: [ date ] [ thread ]
Date: 25 Oct 2004 05:35:33 -0700
> I'm not quite sure why you say abstraction is the only way to predict
> the future. However, since the memory-prediction system needs to
> generalize, it helps to send data to it which is abstracted already.
> Every real image is slightly or greatly different from the others, so
> the common features are what's important.
My 2c: abstaction, generalization, & pattern recognition is the same
process,- discovery of common (invariant) features. Generality is
range*precision of the commonality, what I call accumulated match,
which is the only possible predictor of future match (a quantifier of
induction). Again, many different words that really mean the same
thing.
- Next message: David Longley: "Re: Three Dog Problems"
- Previous message: Glen M. Sizemore: "Re: Hawkins ideas on building AI's"
- In reply to: dan michaels: "Re: Hawkins ideas on building AI's"
- Next in thread: Eray Ozkural exa: "Re: Hawkins ideas on building AI's"
- Reply: Eray Ozkural exa: "Re: Hawkins ideas on building AI's"
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|
