On 10/12/07, Mike Tintner <[EMAIL PROTECTED]> wrote:
> Let me be brutally challenging here : the reason you guys are attached to
> purely symbolic models of the world is not because you have any real
> evidence of their being productive (for AGI), but because they're what you
> know how to do. Hence Vlad's "why can't 3D world model be just described
> abstractly.." He doesn't know - he just hopes - that it can. Logically.
> What you need here is not logic but - ahem - evidence {sensory stuff].
As Edward eloquently explained, sensory input can as well be regarded
as symbolic. The only difference for anthropomorphic sensory input is
in that it seems to be close to actual details that need to be
perceived. Vision supplies too much data, so visual perception is
mostly about filtering out unnecessary details, and all useful data
also seems being present there. So it creates an illusion that you
actually need all those useful details supplied directly as part of
experience.
But when trained human understands textual description, this
description doesn't contain all those details - he uses his knowledge
about them. And if he needs additional details in order to understand
the scene, to build a consistent model of it, he can ask for
additional details. These details parameterize relatively simple model
of spacial scenes which enables description of relative positions,
movement and simple physics. Now, this model can be described
abstractly, and all you need after that is to specify models of
various objects in it, plus animations of common actions. All these
things can be gradually refined through high-level description,
although admittedly it would be a tedious process and we'll likely be
better off with some kind of fovea-limited dynamic vision.
Generation of such abstract-description-based scenes can be a tedious
process at start, involving calculations 'by hand' on part of AGI, but
gradually through introduction of intermediate concepts this process
will become more intuitive and finally world model will be as flexible
as one directly obtained from vision. So, abstraction-based model can
be used as surrogate vision supply, results of which can be more
optimally reperceived.
--
Vladimir Nesov mailto:[EMAIL PROTECTED]
-----
This list is sponsored by AGIRI: http://www.agiri.org/email
To unsubscribe or change your options, please go to:
http://v2.listbox.com/member/?member_id=8660244&id_secret=52789112-8cab43
