Atomspace- A neural-symbolic network?

[Vishnu Priya]

Hello All,

 

 i read that, Atomspace is a neural-symbolic network. How? 

or is it only a weighted hypergraph? Could any one please explain?

Thanks,

Vishnu

[Matt Chapman]

It's a weighted, labeled hypergraph, where a "symbol" can be represented by a single node or a network-of-nodes/sub-graph. Ideally, Inference and Hebbian Learning rules allow new "symbols" to form automatically as new nodes are created to connect groups of related nodes. The weights of the nodes are probabilistic truth values.

That's how I know to summarize it; hope it helps. Beware, I'm a hobbyist lurker, not active on the project currently, and I may be completely wrong or significantly misleading, with significant non-zero probability. ;-) Read the wiki for further details and possibly more precision, though beware the possibility of vastly outdated content, so the real source-of-truth is the code on github.

All the Best,

Matt

--

Standard Disclaimer:
Please interpret brevity as me valuing your time, and not as any negative intention.

--
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+unsubscribe@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/cf6e5d9c-6c9d-454a-b2b3-bfb179a05102%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Amirouche Boubekki]

I think the neural part comes from the fact that it use an (hyper) graph of probabilities like explained by Matt. The symbolic part stems from the fact that nodes in the graph are labeled which strings which form a programming language that can take advantage of FOL. My understanding is that atomspace is the paper on which you write AGI programs. A regular programming language is coded in text file, OpenCog is coded in atomspace. Or maybe I am misled. I am just a lurker too.

On Thu, May 4, 2017 at 8:20 PM Matt Chapman <ma...@ninjitsuweb.com> wrote:

It's a weighted, labeled hypergraph, where a "symbol" can be represented by a single node or a network-of-nodes/sub-graph. Ideally, Inference and Hebbian Learning rules allow new "symbols" to form automatically as new nodes are created to connect groups of related nodes. The weights of the nodes are probabilistic truth values.

That's how I know to summarize it; hope it helps. Beware, I'm a hobbyist lurker, not active on the project currently, and I may be completely wrong or significantly misleading, with significant non-zero probability. ;-) Read the wiki for further details and possibly more precision, though beware the possibility of vastly outdated content, so the real source-of-truth is the code on github.

All the Best,

Matt

--

Standard Disclaimer:
Please interpret brevity as me valuing your time, and not as any negative intention.

On Wed, May 3, 2017 at 1:36 AM, Vishnu Priya <vishnup...@gmail.com> wrote:

Hello All,

 

 i read that, Atomspace is a neural-symbolic network. How? 

or is it only a weighted hypergraph? Could any one please explain?

Thanks,

Vishnu

--
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/cf6e5d9c-6c9d-454a-b2b3-bfb179a05102%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--
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/CAPE4pjAFPXdKTyCbCe9r8Zi%2B1w-8sa7T6b20n69Bj9_%3Dg%2ByNTw%40mail.gmail.com.