This is the AI stuff from the Palm guy... correct? I have looked at
this. And yes, Jff Hawkins (co-author of On Intelligence) gets it!
Compression in this context means the reduction towards general or
prototypical forms that are then stored with modifiers that enumerate
the delta between a specific incarnation and that prototype.
Randall
On Feb 1, 2009, at 3:02 PM, Mark Wieder wrote:
Randall-
Sunday, February 1, 2009, 12:40:58 PM, you wrote:
The problem with fractal compression is that it like all
compression schemes is usually applied indiscriminately to a whole
file. A scheme that works best for some data doesnt always work
best for others. The real breakthrough will involve an entropy
metric that can be used to self optimize a scheme to regions and
others other other regions... Setting up a topograhic mapping of
compressions as directed by the morphology of that data. Do that
and you can selectively and accurately reduce an image (or any other
data set) to more general terms. For instance, if a filter were
able to extract obects (humans, plants, buildings, furnature,
infrastructure, land use, animals, equipment) into semantic
primitives, it sould describe and store reality the way our brains
do while we sleep. Achieving million-to-one compression ratios.
Sounds like you may be interested in what Numenta's up to...
www.numenta.com
--
-Mark Wieder
[email protected]
_______________________________________________
use-revolution mailing list
[email protected]
Please visit this url to subscribe, unsubscribe and manage your
subscription preferences:
http://lists.runrev.com/mailman/listinfo/use-revolution
_______________________________________________
use-revolution mailing list
[email protected]
Please visit this url to subscribe, unsubscribe and manage your subscription
preferences:
http://lists.runrev.com/mailman/listinfo/use-revolution
