how does the concepts of PLN fit into the opencog system

[Apil Tamang]

Been reading through the PLN book, largely because I really couldn't get hands-on on any examples or tutorials exploring various concepts of opencog in a more streamlined way. I am starting to get the impression that much of opencog is really just a framework for doing advanced machine learning work. Let me know how true that is.

Not really griping here either. PLN has been a good read. A good insight to have at this point is how PLN is relevant to the opencog framework. This might help me understand how PLN is used in a more practical context. The wiki page gives a little, it is far from what I'd call a satisfactory exposition. Where is the integration between concepts form PLN and Opencog itself ? Does PLN really just extend opencog, for that matter?

Thanks for any feedback.

[Ben Goertzel]

Yes, there are not yet any systematic hands-on tutorials, sorry...

OpenCog is not just a toolkit or framework. However, it can
certainly be used that way, and has profitably been used that way
sometimes.

If you look at "Engineering General Intelligence", especially volume
2, you will find a comprehensive cognitive architecture aimed at
human-level AGI. The current OpenCog system implements parts of this
design, and also includes some other stuff.... However, the whole
integrated design as described in that book is not yet ready and
usable... only some pieces are there.... So for now, you can use
those existing pieces for something, or you can work on trying to
complete that whole design .. or you can add new stuff out of your own
head ;) ...

In the overall cognitive architecture, PLN can be used in many ways...
it can be used to help figure out how to achieve system goals, but
also for many other kinds of inference...

If you want to work on using OPenCog to control an autonomous agent,
then one option is to join our effort to make OpenCog control the
Hanson Robot heads (and the avatar model thereof that exists in
Blender) .... This is the only current initiative to try to use
OpenCog as an integrated cognitive architecture rather than a
toolkit/framework..

-- Ben G

> --
> You received this message because you are subscribed to the Google Groups
> "opencog" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to opencog+u...@googlegroups.com.
> To post to this group, send email to ope...@googlegroups.com.
> Visit this group at https://groups.google.com/group/opencog.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/opencog/3fab4600-16aa-47e6-98ed-a7aab243eb0a%40googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.



--
Ben Goertzel, PhD
http://goertzel.org

“I tell my students, when you go to these meetings, see what direction
everyone is headed, so you can go in the opposite direction. Don’t
polish the brass on the bandwagon.” – V. S. Ramachandran

[Apil Tamang]

Thanks for the response. I believe that this is what you coined as 'CogPrime', yes !? Do you guys call it 'OpenCog' these days, or maybe just the 'Head', as ominous as the latter sounds :)

It'd be nice to try and join the initiative on the Hanson robotic head. I'm just not sure where to start from... yet.

Are there any planned remote interactive session(s) during the december hackathon event, by any chance?

[Ben Goertzel]

HEAD is the Hanson Environment for Application Development, which is a
separate body of code used to simulate and control the robot heads,
and also to interface between the robot heads and OpenCog

The AGI design is called CogPrime or PrimeAGI ... and is distinct
conceptually from the OpenCog software platform... but these
distinctions are often fudgy in practice...

I am thinking to do an interactive session for remote developers
sometime later in in December, rather than during the F2F Cogathon ...
I think it's a mistake to mix the two things up too much...

> https://groups.google.com/d/msgid/opencog/a0334333-9605-481b-bed6-73034633081d%40googlegroups.com.