> > KNOWLEDGE EXCHANGE
> > ==================
> > This is a more ambitious feature. I want to create a protocol
> > for exchanging knowledge that are learned from different robots.
> > For example one person can train his robot to recognize fruits
> > and another trains his for faces. Then a web server may provide
> > tools to maintain a database of such "knowledge" files as well
> > as tools to merge them, etc.
>
>
> I don't see how this is generally possible, short of playing a source
> data log forward.  I mean, I can merge and share the knowledge of MY
> systems(1), but any representation I use is reflective of the
> assumptions and theory of my model and may have no possible or easy or
> tractable translation into the representation of of another model.  This
> is not an easy problem to work around in a really functional way.  The
> capability to load the "brain" of one system into another is very useful
>  and doable, but it won't translate well outside of a single
> system design.

Agreed.  Loading knowledge from one Novamente to another is easy, but
loading knowledge from a Novamente instance to one of James's Ai systems
isn't going to be do-able in any simple way.  Basically, the different Ai
systems would have to explicitly be taught a common language.

So, the problem here really should be framed as: Create a LANGUAGE that is
appropriate for communication among different AI systems, rather than among
humans.

My suggestion for this language is a non-innovative one: Just use predicate
logic.  Predicate logic sucks as an internal representation for AGI systems,
but it's good as a lingua franca between AI systems, as it lacks the
ambiguities imposed by the linearity of syntax of human languages.  SUO-KIF
is a standardization of predicate logic for AI purposes, which is an IEEE
standard and is reasonable though I don't personally love all of it.

In order to use SUO-KIF to communicate knowledge about perceptions and
actions, one has to define a set of primitive predicates describing atomic
perceptions and actions.  It shouldn't be hard to do *this* in a
standardized way for each particular sensor or actuator, though.

I am open to suggestions besides predicate logic, or besides the SUO-KIF
representation of predicate logic, for this purpose.

I note that for many AI systems, the expression of their internal
reprsentations in SUO-KIF may be extremely cumbersome.  In fact this is the
case for many aspects of Novamente's internal representations.  However, I'm
not sure there is a real solution to this issue, due to the large
differences in internal representation among various Ai systems.
>
> I expect when AGI-ish technologies start to take off, there will be huge
> trade and commerce in loadable modules that contain certain classes of
> experience and knowledge.

Agreed...


-- Ben G


-------
To unsubscribe, change your address, or temporarily deactivate your subscription, 
please go to http://v2.listbox.com/member/[EMAIL PROTECTED]

Reply via email to